australian journal of management...
TRANSCRIPT
http://aum.sagepub.com/Australian Journal of Management
http://aum.sagepub.com/content/36/1/15The online version of this article can be found at:
DOI: 10.1177/0312896211401682
2011 36: 15Australian Journal of ManagementChristopher S Armstrong, Antonio Davila, George Foster and John RM HandMarket-to-revenue multiples in public and private capital markets
Published by:
http://www.sagepublications.com
On behalf of:
Australian School of Business
can be found at:Australian Journal of ManagementAdditional services and information for
http://aum.sagepub.com/cgi/alertsEmail Alerts:
http://aum.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://aum.sagepub.com/content/36/1/15.refs.htmlCitations:
What is This?
- Apr 8, 2011Version of Record >>
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Article
Australian Journal of Management36(1) 15–57
© The Author(s) 2011Reprints and permission: sagepub.
co.uk/journalsPermissions.navDOI: 10.1177/0312896211401682
aum.sagepub.com
Market-to-revenue multiples in public and private capital markets
Christopher S ArmstrongThe Wharton School, University of Pennsylvania, USA
Antonio Davila IESE, Spain
George Foster Graduate School of Business, Stanford University, USA
John RM HandKenan-Flagler Business School, University of North Carolina, USA
AbstractThe behavior and determinants of market-to-revenue ratios in public and private capital markets is examined. Three samples are analysed: (1) all publicly traded stocks listed at some time on the New York Stock Exchange/American Stock Exchange/National Association of Securities Dealers Automated Quotation System in the 1980–2004 period; (2) sample of over 300 so-called ‘internet companies’ in the 1996–2004 period; and (3) over 5500 privately held venture capital-backed companies in the 1992–2004 period. Both company size and the most recent revenue growth rate are found to explain significant variation across companies in their market-to-revenue multiples — smaller companies and companies with higher recent revenue growth rates have higher multiples. We also document how the capital market appears to use a broad-based information set when setting market-to-revenue multiples for companies with negative revenue growth rates — transitory revenue growth components appear to be identified (in a probabilistic sense) by the capital market. Contrary to much anecdotal comment, we present evidence that the capital market behaved directionally along the lines predicted by capital market theory in the pricing of internet stocks in the 1996–2004 period.
Keywordsinternet companies, market-to-revenue multiple, price/sales ratio, private companies, public companies, transitory revenue, venture capital
Corresponding author:George Foster, Wattis Professor of Management and Dhirubhai Ambani Faculty Fellow in Entrepreneurship, Graduate School of Business, Stanford University, USA. Email: [email protected]
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
16 Australian Journal of Management 36(1)
1. Introduction
The market capitalization to revenue ratio (sometimes called price/sales ratio) is of central interest in many areas of capital market investment analysis and research. These areas include fundamen-tal analysis, acquisition analysis, and examination of returns from quantitative investment trading strategies.1 Revenue forecasts now are part of most security analyst reports.2 This paper provides an extensive analysis of market capitalization to revenue multiples (hereafter MKT
t/REV
t) and
their drivers. We examine MKTt/REV
t multiples for three samples, each of which is of much inter-
est in its own right. Sample one includes all publicly traded stocks listed at some time on the New York Stock Exchange (NYSE), American Stock Exchange (AMEX), or National Association of Securities Dealers Automated Quotation System (NASDAQ) over the 1980–2004 period. Sample two comprises roughly 300 publicly traded ‘internet companies;’ the sample has most of its obser-vations in the 1996–2004 period. Sample three comprises over 5500 privately held, venture-backed companies from primarily the 1992–2004 period. Comparisons across the three samples strengthens the reliability of the inferences we draw about the behavior and determinants of MKT
t/
REVt multiples.
Key findings in this paper include: (1) company size is a significant predictor of MKTt/REV
t ratios, with smaller companies having higher MKT
t/REV
t ratios; (2) past revenue growth is a sig-
nificant predictor of the level of the MKTt/REV
t ratio; and (3) the capital market takes into account
transitory components of revenue when setting MKTt/REV
t ratios. Underlying factors explaining
these findings are examined. The use of data pre, during, and post the 1998–2000 period also enables us to explore factors associated with the so-called ‘internet bubble’ period. Our results are consistent with the capital market behaving in this period more rationally along the predictions of capital market theory than many observers are willing to recognize.
2. Importance of research and literature reviewThe findings in this paper have relevance for multiple research areas, including capital market valuation research, management of reported financial numbers research, capital market 1998–2001 ‘bubble’ research, and early-stage company research.
2.1 Capital market valuation researchMuch valuation research using market multiples has emphasized market (price)-to-earnings or market-to-book value. MKT
t/REV
t analysis is also important in valuation research. Analysis of
market-to-revenue multiples can be motivated by both the revenue variable itself being an informa-tive signal about future value-relevant parameters and by revenues being a ‘substitute’ variable for situations in which negative earnings or book value are encountered.
A growing number of studies have documented the differential or incremental information con-tent of revenue vis-à-vis other variables. Ertimur et al. (2003) report that a dollar of surprise in rev-enues has higher capital market information content than a dollar of surprise in expenses. Hypothesized explanations include a persistence explanation, a homogeneity explanation, and an earnings management explanation. Berger (2003) argues that the persistence explanation – ‘inves-tors react more strongly to a (revenue) surprise because these are less transitory than expense sur-prises’ (p.213) – is the most credible explanation of the three proposed. Ghosh et al. (2005), using an earnings response coefficient (ERC) methodology, report that ‘relative to the cost reduction firms, earnings for the revenue-growth firms are more persistent… revenue-supported earnings growth is likely to be more sustainable because revenue is the key value driver and its growth often reflects the
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 17
underlying product differentiation strategy’ (p.19). Keung (2010) reports incremental information content to analyst revenue forecasts when they simultaneously make changes to their earnings and revenue forecasts. The research in this paper provides systematic evidence to support the revenue persistence/momentum argument. We also document across multiple samples that companies with higher positive revenue growth in the most recent year have higher MKT
t/REV
t multiples.
Many valuation studies on market multiples, by their choice of variable(s) to examine, exclude sizable numbers of interesting and important observations. For example, research on market (price)-to-earnings multiples invariably deletes observations with non-positive earnings. This dele-tion can have sizable impacts on the generalizability of the results. Appendix A of this paper reports the percentages of NYSE, AMEX, and NASDAQ companies with negative net income. The per-centages in Table 1 relate to the ‘high-technology’ subsample (covering the computer hardware, software, telecom, and biotech/pharmaceuticals industries) and the ‘all other’ subsample of com-panies listed on these exchanges.
Table 1. Percentages of companies with negative net income
1985 1990 1995 2000 2004
High-tech companies 30.3% 37.2% 39.1% 56.8% 50.3%All other companies 16.3% 23.1% 17.2% 22.3% 22.7%
Companies listed on NASDAQ (on average, smaller and younger) have a higher percentage of negative net income than companies listed on the NYSE and AMEX. Core et al. (2003), in a study of market-to-book value multiples, report losing 5355 observations (4.7%) from their sample when requiring companies to have positive book value. They note that ‘deleting firms with negative book value of equity removes a greater percentage of young (7.5%) and high-technology (5.6%) firms’ (p.51).
Research that compares the relative ‘performance’ of MKTt/REV
t vis-à-vis other market mul-
tiples highlights the potentially severe deletion that can occur in traditional MKTt/INC
t or MKT
t/
BVt multiple studies. Consider the Liu et al. (2002) study, which adopted a ‘horse-race’ ranking
methodology in which the ability of alternative accounting variables to predict the current stock price was compared. The variables included six historical series (cash flow, cash flow from opera-tions, earnings before interest, taxes, depreciation, and amortization (EBITDA), revenues (sales), earnings, and book value of equity) and two forward-looking series (analyst forecasts of earnings per share (EPS) and long-term growth in EPS). The conclusion was that ‘despite the importance of top-line revenues, its value-relevance is limited until it is matched with expenses’ (pp.137–138). The sample examined had sizable reductions – they excluded ‘firm-years with negative values for any value driver’ and ‘firms not covered by IBES, typically firms with low and medium market capitalization.’ The authors noted that their ‘results may not be descriptive of startup firms report-ing losses and high growth firms with negative operating cash flows’ (p.138).
The research in this paper examines MKTt/REV
t multiples. The observations we delete due to
non-positive denominators are minimal vis-à-vis studies on MKTt/INC
t or MKT
t/BV
t multiples.3
Moreover, we present results for different samples, some of which represent companies with below average representation in previous research on market multiples (namely technology and smaller companies). The privately held venture-backed company sample we examine has only recently been the subject of empirical research. Our results highlight how companies with either negative net income or negative book value can still be the subject of capital market multiple research, as opposed to being ignored in a ‘deleted observations’ subsample.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
18 Australian Journal of Management 36(1)
2.2 ‘Management’ of reported financial numbers
The earnings management literature has examined multiple dimensions, including analysis of the revenue series, various expense series, balance sheet items, and cash flow items. Stubben (2010) examines the ability of revenue and accrual models to detect simulated and actual earnings man-agement. Callen et al. (2004) illustrate research with a focus on the revenue series. They report finding that ‘revenues are value relevant in explaining market value of equity whereas earnings are not significant in explaining the market value of equity for firms reporting negative earnings. Given our assumption that revenue manipulation flows though accounts receivable, we show that firms with a more extreme string of past and anticipated losses report higher accounts receivable to sales ratio’ (p.30). Bowen et al. (2002) study the revenue reporting policies of internet companies. They report that ‘the pressure to seek external financing influences internet managers’ choices to report barter and/or grossed-up revenue.’ The transitory revenue results in Section 8 of this paper provide further insight into how the reported revenue series may differ from that which the capital market views as value relevant.
2.3 Analysis of 1998–2000 capital market ‘bubble’Phrases such as ‘bubble period,’ ‘irrational exuberance,’ and ‘internet euphoria’ have been used to describe the 1998–2000 capital market period. There were major swings in the market capitaliza-tion of many companies in this period. Multiple studies have been undertaken of the valuation of internet companies in this period (see, for example, Hand, 2001, 2003; Trueman et al., 2000, 2001). These studies typically analyse companies in the era when stock prices were increasing or at his-torically ex post high levels. Core et al. (2003) examine the determinants of market-to-book value multiples for several samples (all firms, high-tech, and young firms) in the 1975–1999 period. They conclude that while the regression models’ ‘explanatory power decline in the New Economy subperiod…the regression model’s structure during the New Economy subperiod is not unusual compared to other subperiods’ (p.43).
We analyse the 1980–2004 period, which includes the subsequent dramatic reductions in market capitalization of many stocks, as well as their prior sizable runups. This coverage provides a stron-ger foundation to make inferences about the 1998–2000 period vis-à-vis other periods. Moreover, we examine venture-backed privately held companies as well as publicly traded companies. This sample provides further insight into the so-called ‘bubble period,’ as many public companies with dramatic increases in market capitalizations in the 1998–2000 period came from our venture-backed sample of companies.
2.4 Early-stage company valuation and growthThere is an emerging literature on how early-stage companies grow over time and the factors that affect their valuation. One stream of this literature has focused on the valuation and financial evo-lution of privately held venture capital (VC)-backed companies (see, for example, Armstrong et al., 2006; Davila et al., 2003; Hand, 2005). Another stream of research examines revenues and revenue forecasts of these companies (see, for example, Armstrong et al., 2007). A third stream examines the evolution of the management control systems (including revenue/sales budgeting) of early-stage companies – see, for example, Davila (2005), Davila and Foster (2005, 2007), Davila et al. (2009, 2010), and Sandino (2007). Research is also focusing on the financial statements prepared or released by startup or privately held ventures – see Allee and Yohn (2009) and Cassar (2009). Hand (2011) documents how, for a sample of 1133 venture-backed companies, both financial
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 19
(current year revenues) and non-financial (patents, headcount, etc.) explain one-year-ahead reve-nue forecasts. Extensive evidence on the revenue growth rates of early-stage companies in many countries is provided by Foster et al. (2011). The research in this paper extends this literature by examining factors that explain differences in the revenues of early-stage companies, as well as the drivers of early-stage company valuation.
3. Research hypotheses examinedWe examine the following three hypotheses.
Hypothesis one. Smaller companies have higher MKTt/REV
t ratios. Smaller companies are hypothesized
to have greater potential for future higher revenue growth rates. There is strong evidence supporting hypothesis one. This finding highlights the impact that sample deletion procedures can play in excluding interesting and important observations. Smaller companies, on average, are more likely to have negative net income or negative book value. They are also less likely to have extensive analyst coverage. In our samples, smaller companies are more likely to come from the high-technology sector, where intangible assets are of much interest. Revenues for such companies potentially are an important signal that they are converting ‘investments’ in intangible assets (ideas, innovations, discoveries, development of a sales force, etc.) into products/services commercially valued by customers or partners.
Hypothesis two. Companies with higher recent positive revenue growth rates will have higher MKTt/
REVt ratios than companies with lower recent positive revenue growth rates. Past revenue growth is
hypothesized to have significant momentum (i.e. persistence), such that the higher the most recent year’s revenue growth, the higher the likely future revenue growth. There is strong evidence supporting hypothesis two.
Hypothesis three. Companies with negative reported revenue growth and higher MKTt/REV
t ratios have
a higher likelihood of negative transitory revenue. Transitory revenue is the difference between reported revenue and underlying revenue. The capital market is hypothesized to use a broad information set (cer-tainly larger than just the most recent revenue growth observation) and is able to identify the companies whose current reported revenue contains a higher negative transitory component. There is strong evidence supporting hypothesis three.
4. Samples of companies examinedThree samples of companies are examined in this paper. These three samples were chosen to ensure sizable cross-sectional and time-series differences in MKT
t/REV
t multiples of the compa-
nies examined. They were also chosen to gain insight into the 1998–2000 period, where MKTt/
REVt multiples for large numbers of publicly and privately traded companies reached levels not
previously encountered.
4.1 Sample one: publicly traded companies in selected standard industrial classification industries on the NASDAQ and NYSE/AMEXCompanies of particular interest in our research are those with extreme movements in market capi-talization. Extreme variation in the numerator of the revenue multiple enables the research to probe the magnitude of any associated variation in the denominator of the market-to-revenue multiple. The 1998–2001 period is a recent period with very large variations in market capitalizations.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
20 Australian Journal of Management 36(1)
Focusing on this period, we used the following criteria to identify two subgroups with differing variations in market capitalization.
Firstly, we calculated the sequence of end-of-month aggregate market capitalizations from January 1998 to December 2001 of all companies on the NASDAQ, AMEX, and NYSE exchanges in every three-digit standard industrial classification (SIC) code. For each three-digit SIC industry, we then found the following:
(1) high value of that three-digit SIC industry group’s market capitalization;(2) low value before the date of the high value but after January, 1998; and(3) low value after the date of the high value but before December 2001.
Secondly, we then calculated the ratio of high-to-low market capitalizations (a) before the peak (from (1) and (2)), and (b) after the peak (from (2) and (3)). The average of the two ratios from (a) and (b) measures the relative increase and decrease in aggregate value of the three-digit SIC industry group during the 1998–2001 period. We next ranked all SIC industries using the average ratio and chose the top six industries with a peak individual aggregate market capitalization of at least $1 trillion. These six SIC industries we put into four industry groups – computer hardware (SIC codes 357, 366, and 367), computer software (737), telecommunications (481), and biotechnology/pharmaceutical (283).4
Table 2(a) reports the aggregate market capitalization information (as of 1 January 1998, 10 March 2000, and 31 December 2001) and the number of companies (as of 10 March 2000) for the following groups:5
I. all companies;II. pooled selected SIC industries (II.A + II.B + II.C + II.D); II.A. computer hardware (357, 366, 367); II.B. computer software (737); II.C. telecommunications (481); II.D. biotechnology/pharmaceuticals (283);III. all other companies (in I but not in II).
At the 10 March 2000 peak, our four selected industry groups comprised approximately 50% of the total market capitalization of the NYSE/AMEX/NASDAQ. NASDAQ firms in Group II made up 57.6% of the total NASDAQ market capitalization. NYSE/AMEX firms in Group II made up 42.4% of NYSE/AMEX total market capitalization.
Although the sample one selection criteria focused only on the 1998–2001 period, it is interest-ing to note that our selected SIC industries subsample has differed from the all other companies subsample on several key risk-related variables for many years prior to the 1998–2001 period. Appendix A reports that from at least 1985, companies in the selected SIC industries subsample have had (a) higher security market risk (measured by both the beta and standard deviation of security returns), and (b) a higher percentage of companies with negative net income vis-à-vis the all other companies subsample. Some key differences we find between subsamples II and III in the 1998–2001 period are by construction of the research design. However, differences we observe both pre-1998 and post-2001 are not by construction.
4.2. Sample two: publicly traded internet companiesIn the late 1990s and early 2000s, the phrase ‘internet company’ was used to describe many new companies where the internet played a key role in user engagement or revenue yield. Revenue
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 21
rather than net income was the more frequently used financial statement item in valuation analysis of these so-called internet companies. This was in part due to most such companies not having positive net income, especially in the 1998–2001 period. Sample two comprises publicly traded internet companies drawn from a database on Jay Ritter’s website (used in Loughran and Ritter,
Table 2. Summary statistics on three samples: sample one (all NYSE/AMEX/NASDAQ public companies), sample two (internet public companies), and sample three (venture-backed private companies)
(a)
Sample one - NYSE/AMEX/NASDAQ companies (public)
Aggregate market capitalization ($ billions) No. of companies
1/1/1998 1 3/10/2000 1 12/31/2001 3/10/2000
1. All companies $11,051 $18,398 $14,212 8401II. Pooled selected SIC industries 2849 9262 4593 1932 (357, 366, 367, 737, 481, 283) II.A. Computer hardware
(357,366,367)933 4098 1612 615
II.B. Computer software (737) 550 2504 979 792 II.C. Telecommunications (481) 528 1387 625 190 II.D. Biotech/pharma (283) 839 1273 1377 335III. All other companies 8202 9135 9619 6469
(b)
Sample two - internet companies (Public)
Aggregate market capitalization ($ billions) No. of companies
1/1/1998 3/10/2000 12/31/2001 3/10/2000
I. All internet companies $80 $1630 $279 367II. Pooled selected internet SIC
industries77 1500 256 260
II.A. Internet - computer hardware 57 653 158 21 II.B. Internet - computer software 17 726 72 178 II.C. Internet - telecommunications 2 47 2 23 II.D. Internet - business services 1 75 24 38III. All other internet companies 3 130 23 107
(c)
Sample three - VC Backed companies (Private)
Number of companies
Number with IPO
Internet Coy’s with IPO(c)
I. All Venture one companies 11,910 1262 239 I.A. Software (a) 4811 388 134 I.B. Consumer-business services 2272 127 48 I.C. Communications 1604 195 53 I.D. Biopharmaceuticals 1008 231 0 I.E. Hardware/Equipment (b) 2215 321 4
NSYE: New York Stock Exchange, AMEX: American Stock Exchange, NASDAQ: National Association of Securities Dealers Automated Quotation System, SIC: standard industrial classification, IPO: initial public offer(a) Comprises the ‘software’ and ‘information services’ industry segments of VentureOne. (b) Comprises the ‘medical devices,’ ‘electronics,’ and ‘semiconductors’ industry segments of VentureOne. (c) The internet company list (taken from Jay Ritter’s website) is combined with the VentureOne listing to identify the VentureOne internet companies with an IPO.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
22 Australian Journal of Management 36(1)
2004). The database is built from a merging of ‘Internet identifications of Thompson Financial Securities Data, Dealogic, and IPOMonitor.com’ (p.1).6 To facilitate comparisons with our SIC industry analysis, we cross-classified the internet sample with our SIC-based groupings. The over-lap is strongest for computer software, computer hardware, and telecommunications industries. However, one other three-digit SIC industry (738 – miscellaneous business services) accounted for over 10 % of the internet sample (as of 10 March 2001).
Table 2(b) lists the aggregate market capitalizations (as of 1 January 1998, 10 March 2000, and 31 December 2001) and the number of companies (as of 10 March 2000) for the following groups:
I. all internet companies;II. pooled selected internet-SIC industries (II.A + II.B + II.C + II.D); II.A. internet-computer hardware; II.B. internet-computer software; II.C. internet-telecommunications; II.D. internet-(miscellaneous) business services;III. all other internet companies (in I but not in II).
The ‘all other internet companies’ group (III) consists of internet companies from many SIC industry groupings. However, no SIC industry group in (III) has more than 4% of the all internet company group (I).
The internet sample is dominated by companies that went public after 1995.7 In contrast, sample one includes companies of many different initial public offer (IPO) and age vintages. The internet sample is almost exclusively traded on the NASDAQ. On 10 March 2000, 96.2% of the total com-panies in sample two were listed on the NASDAQ. The NASDAQ internet companies made up 97.8% of the total NASDAQ/NYSE/AMEX internet market capitalization on that date.
4.3 Sample three: privately held venture-backed companiesRevenues are a pivotal variable for valuing privately held venture-backed companies. VC exists, in part, to finance early-stage companies whose rapid growth aspirations often result in their having negative operating cash flow (and often negative income) in their early years. Our sample three draws on a Dow Jones VentureSource (previously called VentureOne) database. VentureSource is a commercial organization that collects and sells information about venture-backed companies and their investors. The focus is on each company up to the time of an IPO, a trade sale, or some other exit. For each company included in its database, information on private financing rounds (such as dates of funding rounds, amounts raised, and pre-money valuations) as well as details about the com-pany’s management and investors is available. Several financial statement-based numbers (revenue and net income) are also included for a subset of these companies. The database is at its most com-prehensive from the early 1990s onwards. Data are provided to VentureSource by companies and their investors on a voluntary basis. Where possible, VentureSource uses additional sources to verify the reported numbers (such as obtaining pre-money valuation numbers from the company itself, from individual investors, tracking business press reports on the company’s financing, and publicly avail-able regulatory reports, such as S-1 filings with the Securities & Exchange Commission (SEC)).
VentureSource provides its own industry classifications (16 in total) for the companies in its database. The top eight industry classifications cover 86.5% of the 13,765 companies in the data-base. We use these eight industries as our sample three, grouping them into five broader industries:
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 23
I. all companies in sample three; I.A software; I.B consumer-business services; I.C communications; I.D biopharmaceuticals; I.E hardware/equipment.
Table 2(c) summarizes the composition of these VentureSource industry groups. Of the 11,910 companies in Table 2(c), 1262 had gone IPO by March 2005. We also report the number of these IPO companies that were classified as internet companies using the sample two internet company listing (from Jay Ritter’s website). Approximately 18.9% (or 239 companies) of this 1262 IPO sample became publicly listed internet companies, with most (92.8%) of these having their IPO between 1996 and 2000. At the time of their public listing, the majority (55.25%) of the VentureSource companies with IPOs are classified into the six three-digit SIC code industries that comprise our selected SIC industry group in sample one.
5. Time-series and cross-sectional distribution of MKTt/REV
t multiples
Figure 1 presents time-series plots from 1980 to 2004 for the MKTt/REV
t (the ratio of market capi-
talization of common equity to the most recent annual revenue) multiple for samples one to three. Each plot has the 90th, 70th, 50th, 30th, and 10th percentiles of the annual cross-sectional distribution. These percentiles are computed on a year-by-year basis. Sample one has two subsamples. Both have similar scales (0–100) in Figure 1 to facilitate comparability. The selected SIC industries subsample was selected based on high market capitalization variation in the 1998–2001 period vis-à-vis the all other companies subsample. There are dramatic MKT
t/REV
t multiple differences
across these two subsamples in Figure 1. The 90th and 70th percentiles of the selected SIC indus-tries subsample are consistently higher than that of the comparable deciles for the all other compa-nies subsample. The MKT
t/REV
t plots for samples two and three are scaled (0–350) higher than
for sample one to reflect the sizably higher 90th MKTt/REV
t percentile values, especially in the
1999–2000 period. Specific values for the 90th percentile across the different samples for selected years are given in Table 3.
Table 3. Specific values of MKTt/REV
t for the 90th percentile
1985 1995 1998 1999 2000 2001 2002 2004
Sample one: NASDAQ/NYSE/AMEX Selected SIC industries 6.6 20.2 18.6 80.3 46.1 27.4 13.9 41.7 All other companies 5.1 5.1 4.6 5.9 6.1 5.4 4.4 7.2Sample two: internet companies - - 85.6 233.1 33.5 9.1 5.5 11.6Sample three: venture-backed companies - 78.3 142.5 161.0 332.5 149.8 233.4 100.0
NSYE: New York Stock Exchange, AMEX: American Stock Exchange, NASDAQ: National Association of Securities Dealers Automated Quotation System, SIC: standard industrial classification
In subsequent sections of this paper, we examine hypotheses that pertain to these time-series or cross-sectional differences in MKT
t/REV
t multiples.8
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
24 Australian Journal of Management 36(1)
6. Hypothesis one: company size and MKTt/REV
t ratios
Hypothesis one is that smaller companies have higher MKTt/REV
t ratios. Theoretical valuation
models often include a growth term. For example, in the classic Miller–Modigliani (1961) frame-work, growth was defined as investments with returns greater than their cost of capital – see Fama and Miller (1972). Much subsequent valuation research has used the Black–Scholes (1973) option-theoretic framework to incorporate growth options into equity valuation. Hypothesis one rests on smaller companies providing higher growth potential than larger companies. At an extreme level, high compound growth rates over extended periods will constrain larger companies before smaller companies, due to theoretical limits imposed by the size of the market, industry, or economy in which the company operates. For any innovative product that creates a given new market demand, the relative effect on a company’s revenue or net income will be greater for a smaller company than for a larger company. Smaller companies may also encounter fewer regulatory obstacles in their early growth strategies.
Figure 2 shows the 90th, 70th, 50th, 30th, and 10th percentiles of distribution for the MKTt/REV
t
ratio for five company size categories. These five categories (in $ millions of revenues) and the percentage of observations in each category for the three samples we examine are given in Table 4.
Sample OneSelected SIC Industries (Public)a
Sample OneAll Other Companies (Public)
Sample TwoInternet Companies (Public)
Sample ThreeVC Backed Companies (Private)
0102030405060708090
100
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0102030405060708090
100
1980 1985 1990 1995 2000 2005
0
50
100
150
200
250
300
350
1980 1985 1990 1995 2000 20050
50
100
150
200
250
300
350
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
90th70th50th30th10th
90th70th50th30th10th
Figure 1. Market capitalization to revenue multiple percentiles (90th, 70th, 50th, 30th, and 10th) for sample one (all New York Stock Exchange/American Stock Exchange/National Association of Securities Dealers Automated Quotation System public companies), sample two (internet public companies) and sample three (venture-backed private companies): 1980–2004. SIC: standard industrial classification. aComprises companies in hardware, software, telecom, and biotech/pharmaceuticals industries.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 25
Selected SIC Industries(Public)
All Other Companies(Public)
Rev
< $
5M$5
M <
Rev
< $
25M
$25M
< R
ev <
$10
0M$1
00M
< R
ev <
$1,
000M
Rev
> $
1,00
0M
(a)
0
50
100
150
200
250
300
350
400
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
50
100
150
200
250
300
350
400
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
5
10
15
20
25
30
35
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
5
10
15
20
25
30
35
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
2
4
6
8
10
12
14
16
18
20
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
2
4
6
8
10
12
14
16
18
20
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
2
4
6
8
10
12
14
16
18
20
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
2
4
6
8
10
12
14
16
18
20
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
-1
1
3
5
7
9
11
13
15
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
-1
1
3
5
7
9
11
13
15
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
(Continued)
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
26 Australian Journal of Management 36(1)
Sample Two – Internet(Public)
Sample Three –VCBacked (Private)
Rev
< $
5M$5
M <
Rev
< $
25M
$25M
< R
ev <
$10
0M$1
00M
< R
ev <
$1,
000M
Rev
> $
1,00
0M
(b)
0
500
1000
1500
2000
2500
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
500
1000
1500
2000
2500
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
50
100
150
200
250
300
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
50
100
150
200
250
300
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
20
40
60
80
100
120
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
20
40
60
80
100
120
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
10
20
30
40
50
60
70
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
10
20
30
40
50
60
70
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
5
10
15
20
25
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
0
5
10
15
20
25
1980 1985 1990 1995 2000 2005
90th70th50th30th10th
Figure 2. (a) Annual market capitalization-to-revenue multiple percentiles for sample one companies in selected SIC industries for various company size categories: 1980–2004. (b) Annual Market capitalization-to-revenue multiple percentiles for sample two and sample three companies for various company size categories: 1980–2004. SIC: standard industrial classification, VC: venture capital.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 27
Not surprisingly, sample three (VC-backed private companies) contains the highest percentage of observations in each of the two smallest company size categories ($0–5 million and $5–25 million in revenue) and the lowest percentage in the largest company size category (>$1000 million in revenue). The all other companies subsample of sample one has the lowest percentage of observa-tions in the two smallest categories and the highest percentage in the largest category. The selected SIC industries subsample has 41.2% of observations in the $0–25 million revenue size range com-pared to 27.2% of observations for the all other companies subsample of sample one.
Figure 2(a) shows the sample one MKTt/REV
t distribution over the 1980–2004 period for the
five company size categories. The vertical MKTt/REV
t scalings on the five size categories are as
follows: $0–5 million (scale of 0–400); $5–25 million (0–35); $25–100 million (0–20); $100–1000 million (0–20); >$1000 million (0–15). The 90th percentile MKT
t/REV
t value for the smallest com-
pany group ($0–5 million) is consistently, over time, higher than the 90th percentile MKTt/REV
t
value of the other four company size categories for both the selected SIC industries subsample and the all other companies subsample. Figure 2(b) shows company size breakdowns for samples two and three. Results are shown only for a subset of years and a subset of company size categories due to either zero or minimal observations in some years or company size categories. For example, it is only since (approximately) 1996 that a sizable number of internet companies have been publicly listed. Very few VC-backed companies with over $100 million of revenues have remained private stand-alone companies. The vertical MKT
t/REV
t scalings in Figure 2(b) on the five size categories
are $0–5 million (scale of 0–2500); $5–25 million (0–300); $25–100 million (0–120); $100–1000 million (0–70); >$1000 (0–25). The most extreme MKT
t/REV
t values plotted in Figure 2(b) are
those for the smallest company size category for both samples two and three.In Section 9 of this paper we include company size (revenues) as an independent variable in a
multivariate regression analysis with MKTt/REV
t as the dependent variable. The company size
variable consistently has a negative coefficient that is highly statistically significant across all samples examined. In the next section of this paper we explore the recent revenue growth hypoth-esis and highlight how company size also is central to understanding the relative effect of recent revenue growth on MKT
t/REV
t ratios.
7. Hypothesis two: recent revenue growth and MKTt/REV
t ratios
Hypothesis two is that companies with higher recent positive revenue growth rates will have higher MKT
t/REV
t ratios than companies with lower recent positive revenue growth rates. Its rationale is
that the most recent revenue growth is a value-relevant predictor of future revenue growth or other value-relevant variables. The early literature on the time-series properties of accounting numbers focused on earnings. The key conclusion from multiple studies was that the first differenced earnings
Table 4. The five company size categories and the percentage of observations in each category
I $0–5 II $5–25 III $25–100 IV $100–1000 V >$1000
Sample one: NASDAQ/NYSE/AMEX Selected SIC industries 17.5% 23.7% 24.8% 23.8% 10.2% All other companies 10.8% 16.4% 21.6% 34.1% 17.4%Sample two: internet companies 10.5% 27.3% 31.7% 26.2% 4.3%Sample three: venture-backed companies 32.8% 40.1% 21.1% 5.9% 0.1%
NSYE: New York Stock Exchange, AMEX: American Stock Exchange, NASDAQ: National Association of Securities Dealers Automated Quotation System, SIC: standard industrial classification
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
28 Australian Journal of Management 36(1)
series was either serially uncorrelated or in some studies exhibited statistically significant negative serial correlation – see, for example, Ball and Watts (1972).9 In contrast, the revenue series shows statistically significant evidence of positive serial correlation. We computed autocorrelations for lags 1–4 for the first differenced annual revenue series for companies in our two groups for sample one.10 The time period is 1980–2004. Table 5 shows the results. For comparison purposes, we also report results for the operating income series and the income before extraordinary items series. The selected SIC industries group in sample one has over 20,000 observations, while the all other industries group has over 90,000 observations. Table 5 highlights the statistically significant evidence of positive autocorrelation for the first differenced annual revenue series for both sample one groups. Moreover, the selected SIC industries group (r
1 = 0.36; r
2 = 0.18; r
3 = 0.16; r
4 = 0.17) shows stronger positive
autocorrelations than the all other companies group (r1
= 0.18; r2
= 0.02; r3
= 0.07; r4
= 0.09). The positive autocorrelations for revenues in Table 5 contrast sharply with the minimal to negative auto-correlations for both the operating income series and the income before extraordinary items series.
We now explore further evidence pertaining to hypothesis two – companies with higher recent positive revenue growth rates will have higher MKT
t/REV
t ratios. In Section 9 we present multi-
variate regression results that strongly support this hypothesis. This section of our paper highlights key factors that warrant recognition in quantifying the importance of past revenue growth in MKT
t/
REVt analysis. Annual revenue growth is computed as
Re
Re
venue
venuet
t −−
1
1 × 100%
For a subset of observations there is information for revenuet, but not for revenue
t–1 in our data-
bases. For samples one and two, this is primarily due to newly listed companies. Sample three has a subset of missing observations due to the VentureSource data not always having a full sequence of annual revenue numbers. Note that where the private company in Sample three starts in year t, there is no year t – 1 annual revenue number by definition. The percentage of observations that fall in the positive revenue growth (or no change), negative revenue growth, and missing observations categories are given in Table 6.
The ‘missing observation’ category appears to be not a random sample and is of considerable interest in any study on market-to-revenue multiples or revenue growth
Table 5. Mean autocorrelations for first differenced annual financial series for sample one (NYSE/AMEX/NASDAQ publicly traded companies): 1980–2004
r1
r2
r3
r4
Revenues Selected SIC industries 0.36 0.18 0.16 0.17 All other companies 0.18 0.02 0.07 0.09Operating income Selected SIC industries 0.03 –0.06 0.03 0.01 All other companies –0.10 –0.01 –0.01 0.01Income before extraordinary items Selected SIC industries –0.25 –0.07 –0.10 0.02 All other companies –0.27 –0.04 –0.07 0.02
NSYE: New York Stock Exchange, AMEX: American Stock Exchange, NASDAQ: National Association of Securities Dealers Automated Quotation System, SIC: standard industrial classification
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 29
7.1. Annual revenue growth rates, industry/sector, and time period
Cross-sectional distributions of the annual revenue growth rate from 1980 to 2004 are presented in Figure 3 for the selected SIC industries (a) and all other companies (b) subsamples of sample one. Several patterns are observable in Figure 3. Firstly, the 90th percentile for the selected SIC industries group consistently exceeds the 90th percentile for the all other companies group over the 1980–2004 period. The selected SIC industries has consistently had higher revenue growth in the upper tail of its distribution. Secondly, there is consistently larger revenue growth variability for the selected SIC industries group than the all other companies group. One measure of variability in revenue growth is the 90th–10th interpercentile range. Figure 3(c) plots this interpercentile range for the two groups over the 1980–2004 period. For 1999 and 2000, the interpercentile range had values of 266% and 305%, respectively, which is well above the range outside of the 1998–2000 period. Every year from 1980 to 2004, the interpercentile range for the selected SIC industries is larger. Recall that the indus-tries in the two subsamples were chosen using only 1998–2001 information. The results in Figure 3 and Appendix A highlight how the selected SIC industries exhibited higher variability and higher risk in both capital market and fundamental measures long before the 1998–2001 period.11
The MKTt/REV
t multiples in Figure 1 for sample two (the internet sector) typically exceed
those for the selected SIC industries in Figure 1 for comparable years. Figure 4 presents annual revenue growth rates for the 90th, 70th, and 10th percentiles that highlight that sample two like-wise has higher revenue growth rates vis-à-vis sample one for the 90th and 70th percentiles, espe-cially in the 1998–2000 years. The magnitude of the 90th and 70th percentiles for the internet companies (i.e. sample two) during 1998–2000 are well above either of the two sample one sub-groups for comparable years and well above historical revenue growth rates for the sample one subgroups before the advent of internet companies.
Figure 4 also plots annual revenue growth rate deciles for the privately held venture-backed com-panies (sample three). The vertical scales for samples one and two in Figure 4 are identical (90th percentile: scale of 0%–800%; 70th percentile: 0%–300%) to visually highlight the higher annual revenue growth for the internet companies versus the selected SIC industries and for the selected SIC industries versus all other companies. Sample three is scaled differently (90th percentile: scale of 0%–2000%; 70th percentile: 0%–500%) to avoid differences across other samples being visually diminished by the extreme revenue growth rates for the venture-backed private companies. Sample three has even higher annual revenue growth rates than either samples one or two. Note also the dramatic declines in annual revenue growth rates in Figure 4 for the selected SIC industries (in 2001), the internet companies (in 2001), and the venture-backed companies (in 2002 and 2003).
Table 6. The percentage of observations that fall in the positive revenue growth (or no change), negative revenue growth, and missing observations categories
Positive revenue change
Negative revenue change
Missing observation
Total
Sample one: NASDAQ/NYSE/AMEX companies Selected SIC industries 58.2% 28.7% 13.1% 100.0% All other companies 62.6% 27.6% 9.8% 100.0%Sample two: internet companies 48.3% 26.1% 25.6% 100.0%Sample three: venture-backed companies 61.9% 7.2% 30.9% 100.0%
NSYE: New York Stock Exchange, AMEX: American Stock Exchange, NASDAQ: National Association of Securities Dealers Automated Quotation System, SIC: standard industrial classification
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
30 Australian Journal of Management 36(1)
Table 7 shows the distribution (10th, 20th…80th, 90th percentiles) of revenue growth rates for our five different size company categories and for the pooled sample. The data in Table 7 highlights how smaller companies have both higher negative revenue growth rates (see the 10th and 20th percentiles) and higher positive revenue growth rates (see the 80th and 90th percentiles).
(a) Sample one: Selected SIC industries
(b) Sample one: All other companies
(c) Sample one
-100%
-50%
0%
50%
100%
150%
200%
250%
300%
350%
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Annu
al R
even
ue G
row
th
90th Percentile
10th Percentile
-100%
-50%
0%
50%
100%
150%
200%
250%
300%
350%
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Annu
al R
even
ue G
row
th
10th Percentile
90th Percentile
0%
50%
100%
150%
200%
250%
300%
350%
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Inte
rper
cent
ile (9
0th
min
us 1
0th)
Ra
nge
of A
nnua
l Rev
enue
Gro
wth
Selected SIC Industries (II) All Other Companies (III)
Figure 3. Annual revenue growth percentiles (10th, 30th, 50th, 70th, and 90th) and interdecile range (90th – 10th) for publicly traded companies on New York Stock Exchange/American Stock Exchange/National Association of Securities Dealers Automated Quotation System: 1980–2004. (a) Selected standard industrial classification (SIC) industries (hardware, software, telecom, biotech/pharma). (b) All other companies. (c) Annual interpercentile (90th–10th) range.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 31
SA
MP
LE
ON
ES
AM
PL
E T
WO
SA
MP
LE
TH
RE
ES
elec
ted
SIC
Ind
ustr
ies
(Pu
bli
c)A
ll O
ther
Com
pan
ies
(Pu
bli
c)In
tern
etC
omp
anie
s (P
ub
lic)
Ven
ture
-Bac
ked
Com
pan
ies
(Pri
vate
)90thPercentile 70th Percentile 10th Percentile
0%10
0%20
0%30
0%40
0%50
0%60
0%70
0%80
0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
0%10
0%20
0%30
0%40
0%50
0%60
0%70
0%80
0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
0%10
0%20
0%30
0%40
0%50
0%60
0%70
0%80
0%
1996
1997
1998
1999
2000
2001
2002
2003
2004
0%20
0%40
0%60
0%80
0%10
00%
1200
%14
00%
1600
%18
00%
2000
%
1996
1997
1998
1999
2000
2001
2002
2003
2004
0%50%
100%
150%
200%
250%
300%
1996
1998
2000
2002
2004
0%50%
100%
150%
200%
250%
300%
1996
1998
2000
2002
2004
0%50%
100%
150%
200%
250%
300%
1996
1998
2000
2002
2004
0%50%
100%
150%
200%
250%
300%
350%
400%
450%
500%
1996
1998
2000
2002
2004
-100
%-8
0%
-60%
-40%
-20%0%20%
40%
60%
1996
1998
2000
2002
2004
-100
%-8
0%
-60%
-40%
-20%0%20%
40%
60%
1996
1998
2000
2002
2004
-100
%-8
0%
-60%
-40%
-20%0%20%
40%
60%
1996
1998
2000
2002
2004
-100
%-8
0%
-60%
-40%
-20%0%20%
40%
60%
1996
1998
2000
2002
2004
Fig
ure
4. A
nnua
l rev
enue
gro
wth
rat
es fo
r pu
blic
ly-t
rade
d co
mpa
nies
and
pri
vate
ven
ture
-bac
ked
com
pani
es: 1
996–
2004
. Sel
ecte
d st
anda
rd in
dust
rial
cl
assi
ficat
ion
(SIC
) in
dust
ries
ver
sus
all o
ther
com
pani
es v
ersu
s in
tern
et c
ompa
nies
ver
sus
vent
ure-
back
ed c
ompa
nies
.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
32 Australian Journal of Management 36(1)
Tabl
e 7.
Ann
ual r
even
ue g
row
th r
ate
perc
entil
es (
10th
, 20t
h…90
th)
by c
ompa
ny s
ize
cate
gory
($m
illio
ns: I
, $0–
5; II
, $5–
25; I
II,
$25–
100;
IV, $
100–
1000
; >$1
000)
for
sam
ples
one
, tw
o, a
nd t
hree
Sam
ple
one
–sel
ecte
d S
IC in
dust
ries
(pu
blic
): co
mpa
ny s
ize
cate
gori
es
I $0–$
5II $5
–$25
III
$25–
$100
IV
$100
–$10
00V
> $1
000
Pool
ed
All
10th
pct
le-8
5.4%
-32.
8%-2
1.5%
-13.
2%-8
.6%
-33.
4%20
th p
ctle
-56.
7%-1
7.9%
-8.2
%-2
.1%
-0.1
%-1
3.6%
30th
pct
le-3
7.4%
-7.9
%0.
3%4.
5%4.
2%-2
.8%
40th
pct
le-1
9.7%
0.8%
7.5%
10.2
%7.
5%4.
8%50
th p
ctle
-4.3
%8.
4%14
.4%
15.7
%10
.8%
11.4
%60
th p
ctle
10.4
%17
.2%
22.0
%22
.1%
14.3
%18
.8%
70th
pct
le31
.3%
28.4
%31
.9%
31.0
%19
.4%
28.7
%80
th p
ctle
76.3
%46
.0%
46.7
%45
.5%
26.3
%45
.6%
90th
pct
le23
3.2%
94.0
%79
.6%
75.6
%41
.6%
89.3
%no
bs50
9068
7271
8469
2029
6929
,035
Rev
enue
gr
owth
rat
e pe
rcen
tiles
Sam
ple
one
-All
oth
er c
om
pani
es (
publ
ic):
com
pany
siz
e ca
tego
ries
I 0–$5
II
$5–$
25III
$2
5–$1
00IV
$1
00–$
1000
V
> $
1000
Pool
ed
All
20th
pct
le-5
2.5%
-11.
7%-5
.5%
-2.3
%-1
.6%
-6.6
%30
th p
ctle
-31.
3%-3
.8%
0.5%
2.3%
2.0%
-0.2
%40
th p
ctle
-15.
7%2.
1%5.
0%6.
0%5.
1%4.
3%50
th p
ctle
-3.5
%7.
4%9.
4%9.
7%7.
9%8.
4%60
th p
ctle
7.6%
13.6
%14
.4%
13.7
%10
.9%
12.9
%70
th p
ctle
23.6
%21
.8%
21.0
%18
.8%
14.8
%18
.9%
80th
pct
le55
.4%
35.5
%31
.1%
26.4
%20
.6%
28.6
%90
th p
ctle
157.
0%70
.9%
54.1
%43
.0%
33.4
%52
.7%
nobs
13,7
0220
,752
26,9
7443
,222
22,0
4312
6,69
3
Rev
enue
gr
owth
rat
e pe
rcen
tiles
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 33
Sam
ple
two-i
nter
net
com
pani
es (
Pub
lic):
com
pany
siz
e ca
tego
ries
I $0–$
5II $5
–$25
III
$25–
$100
IV
$100
–$10
00V
> $1
000
Pool
ed
All
10th
pct
le-8
6.6%
-38.
0%-2
3.7%
-12.
3%-8
.2%
-28.
8%20
th p
ctle
-51.
6%-1
5.4%
-5.0
%0.
8%0.
1%-8
.0%
30th
pct
le-1
7.4%
5.1%
11.3
%8.
5%5.
9%7.
4%40
th p
ctle
29.0
%34
.8%
26.7
%22
.2%
10.6
%24
.3%
50th
pct
le70
.2%
64.1
%48
.7%
34.1
%13
.7%
47.1
%60
th p
ctle
163.
0%10
9.7%
78.2
%54
.8%
26.0
%77
.3%
70th
pct
le29
5.8%
194.
4%13
0.3%
87.5
%48
.9%
129.
2%80
th p
ctle
684.
9%32
0.3%
224.
6%14
4.8%
68.4
%23
3.8%
90th
pct
le13
59.2
%63
2.5%
448.
5%29
6.8%
90.5
%52
3.6%
nobs
229
596
690
571
9321
79
Rev
enue
gr
owth
rat
e pe
rcen
tiles
Sam
ple
thre
e-ve
ntur
e ca
pita
l-ba
cked
co
mpa
nies
(P
riva
te):
com
pany
siz
e ca
tego
ries
I $0–$
5II $5
–$25
III
$25–
$100
IV
$100
–$10
00V
> $1
000
Pool
ed
All
10th
pct
le-3
3.4%
5.2%
3.4%
1.1%
-2.1
%20
th p
ctle
0.0%
20.0
%12
.7%
6.9%
12.1
%30
th p
ctle
22.9
%32
.8%
20.9
%12
.9%
25.0
%40
th p
ctle
49.7
%47
.0%
30.2
%18
.5%
39.1
%50
th p
ctle
79.3
%64
.2%
40.1
%26
.2%
56.0
%60
th p
ctle
127.
3%89
.4%
52.2
%35
.6%
80.6
%70
th p
ctle
200.
0%11
9.9%
70.6
%45
.0%
118.
5%80
th p
ctle
356.
3%19
7.1%
103.
9%73
.2%
200.
0%90
th p
ctle
771.
6%39
3.1%
195.
7%11
3.7%
422.
2%no
bs16
1019
6510
3428
748
96
Rev
enue
gr
owth
rat
e pe
rcen
tiles
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
34 Australian Journal of Management 36(1)
7.2 MKTt/REV
t multiples and sign/size of revenue growth rates
Other things being equal, it is expected that the capital market favorably values companies with positive recent revenue growth vis-à-vis companies with negative revenue growth. Table 8 reports distribution statistics (10th percentile, 50th percentile, 90th percentile) for the MKT
t/REV
t multi-
ple for three mutually exclusive groups:
A.) companies with positive or zero reported revenue growth in the most recent year;B.) companies with negative reported revenue growth in the most recent year; andC.) companies without the prior or current year’s reported revenue number in the database with
which to compute a revenue growth rate (typically, these are newly listed companies in samples one and two12).
These three groups do not appear to have similar MKTt/REV
t distributions, especially in their
upper tails. The groups with the highest MKTt/REV
t observations in their upper tails are typically
those in either the negative revenue growth group or the missing revenue growth group. This is counter to the expectation noted above.
Consider the MKTt/REV
t distribution comparison between the negative revenue growth group
and the positive revenue growth group. The ratio of the 90th MKTt/REV
t percentile of the negative
revenue growth group to the 90th MKTt/REV
t percentile of the positive revenue growth group
exceeds one for most years. Summary data for this ratio are given in Table 9.For samples one and three, there is a consistent pattern for the 90th MKT
t/REV
t percentile for
the negative revenue growth group exceeding that of the positive revenue growth group. The pat-tern for sample two (internet companies) is less consistent. While the mean ratio of 5.26 is greater than 1, it is heavily impacted by the 1995 observation, where the 90th percentile is 2101.6. The MKT
t/REV
t distribution for the missing revenue observation sample is typically even more posi-
tively skewed than the other two groups in Table 8.Hypothesis three explores the potential role of transitory revenue in explaining the extreme
positive skewness in MKTt/REV
t ratios for the negative revenue growth subsample in Table 8.
8. Hypothesis three: transitory revenue and MKTt/REV
t ratios
Hypothesis three is that companies with negative revenue growth and higher MKTt/REV
t ratios have
a higher likelihood of negative transitory revenue. The concept of transitory revenue is a potential explanation for the Table 8 counter-intuitive result that the median and upper percentiles of the MKT
t/
REVt ratio of companies with negative revenue growth exceed those with positive revenue growth.
Transitory revenue is linked to reported revenue and underlying revenue as follows:
Reported Revenue = Underlying Revenue + Transitory Revenue
Reported revenue is what each company reports in its financial statements. We compute MKTt/
REVt as market capitalization at the end of the fiscal year end divided by the reported revenue for
that fiscal year. Underlying revenue is the (unobservable) variable that the capital market perceives as relevant to capturing the ongoing revenue-generating capacity of a company in a given period. Transitory revenue are components that are noise in inferring underlying revenue from reported revenue.13 Examples of components include random and ‘one-off’ events, activities by manage-ment to shift/misrepresent reported revenues across periods, and the revenues that result from using accounting methods not viewed as appropriate by the capital market.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 35
Tabl
e 8.
MK
Tt/R
EVt M
ultip
le d
istr
ibut
ion
sum
mar
y fo
r su
bsam
ples
form
ed o
n si
gn o
f rev
enue
gro
wth
in y
ear
t
Tota
l O
bs.
Posi
tive
Rev
enue
Gro
wth
Sub
sam
ple
Neg
ativ
e R
even
ue G
row
th S
ubsa
mpl
eM
issi
ng R
even
ue G
row
th S
ubsa
mpl
e
MK
Tt/R
EVt M
ultip
leN
obs
% o
f To
tal
Sam
ple
MK
Tt/R
EVt M
ultip
leN
obs
% o
f To
tal
Sam
ple
MK
Tt/R
EVt M
ultip
leN
obs
% o
f To
tal
Sam
ple
10th
Pe
rc.
50th
Pe
rc.
90th
Pe
rc.
10th
Pe
rc.
50th
Pe
rc.
90th
Pe
rc.
10th
Pe
rc.
50th
Pe
rc.
90th
Pe
rc.
Sam
ple
One
Sel
ecte
d S
IC
1995
1484
0.5
2.3
11.2
906
61.1
%0.
21.
352
.027
518
.5%
0.6
3.6
55.4
303
20.4
%
1998
1962
0.5
2.2
11.8
1165
59.4
%0.
31.
226
.457
429
.3%
0.8
5.2
80.1
223
11.4
%
1999
2154
0.6
3.2
27.3
1182
54.9
%0.
31.
847
.156
426
.2%
1.7
31.6
365.
840
818
.9%
20
0023
130.
42.
826
.413
1356
.8%
0.2
1.5
45.3
582
25.2
%1.
09.
813
2.0
418
18.1
%
2001
2168
0.5
2.8
22.3
1067
49.2
%0.
21.
424
.195
043
.8%
0.7
6.7
145.
715
17.
0%
2004
1726
0.7
2.5
16.0
1146
66.4
%0.
42.
611
4.5
468
27.1
%1.
211
.712
77.2
112
6.5%
Sam
ple
One
All
Oth
er
1995
6524
0.2
1.0
3.8
4453
68.3
%0.
10.
87.
311
9318
.3%
0.2
1.4
14.7
878
13.5
%
1998
6767
0.2
1.1
4.0
4395
64.9
%0.
10.
85.
618
0126
.6%
0.3
1.8
12.3
571
8.4%
19
9967
100.
20.
94.
041
9862
.6%
0.1
0.8
6.7
1865
27.8
%0.
42.
310
4.1
647
9.6%
20
0064
510.
10.
94.
245
2670
.2%
0.0
0.5
10.5
1461
22.6
%0.
42.
683
.946
47.
2%
2001
6075
0.2
1.1
4.8
3394
55.9
%0.
10.
64.
523
2238
.2%
0.3
2.0
132.
335
95.
9%
2004
4966
0.3
1.5
5.7
3642
73.3
%0.
21.
512
.010
6721
.5%
0.7
4.3
330.
325
75.
2%S
ampl
e Tw
o I
nter
net
19
9517
0.4
1.4
8.7
1058
.8%
0.0%
1.0
23.5
2101
.67
41.2
%
1998
691.
66.
128
.033
47.8
%0.
14.
310
.59
13.0
%4.
324
.491
.727
39.1
%
1999
275
2.1
11.9
113.
650
18.2
%0.
62.
037
.211
4.0%
6.8
44.5
282.
321
477
.8%
20
0037
90.
32.
420
.620
754
.6%
0.1
1.3
60.9
154.
0%0.
96.
967
.115
741
.4%
20
0131
90.
42.
410
.417
554
.9%
0.2
1.5
4.9
124
38.9
%0.
63.
612
.720
6.3%
20
0420
20.
83.
011
.013
968
.8%
0.5
2.3
23.4
5929
.2%
1.0
1.9
30.5
42.
0%S
ampl
e T
hree
VC
Pri
vate
Bac
ked
19
9526
20.
73.
817
.718
671
.0%
0.8
13.0
190.
022
8.4%
1.2
26.4
203.
354
20.6
%
1998
274
1.7
10.0
60.0
147
53.6
%2.
114
.521
1.1
155.
5%4.
827
.630
0.0
112
40.9
%
1999
361
2.6
16.5
83.2
238
65.9
%3.
516
.360
0.0
256.
9%2.
134
.364
8.5
9827
.1%
20
0025
43.
524
.216
0.8
150
59.1
%4.
981
.573
9.0
176.
7%3.
856
.219
71.1
8734
.3%
20
0110
71.
55.
345
.351
47.7
%1.
119
.116
49.7
98.
4%1.
410
.039
7.5
4743
.9%
20
0488
1.0
6.9
24.8
5764
.8%
11.2
400.
050
68.8
55.
7%0.
84.
053
8.2
2629
.5%
SIC
: sta
ndar
d in
dust
rial
cla
ssifi
catio
n, V
C: v
entu
re c
apita
l
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
36 Australian Journal of Management 36(1)
8.1 Revenue growth transition matrix analysis
Insight into the role of transitory revenue can be gained by examining the multi-year behavior of revenue growth from portfolios formed using MKT
t/REV
t for (a) positive revenue growth observa-
tions, (b) negative revenue growth observations, and (c) missing observations. We use several steps to construct a transition matrix as follows.
1. For all of the observations in year t, compute and rank the year t revenue growth t/(t – 1) for those observations with available information. Form 10 equally sized portfolios. Portfolio 1 has the lowest (starting with the most negative) revenue growth rate and Portfolio 10 has the highest (finishing with the most positive). An 11th portfolio of observations with missing revenue growth in year t is also formed.
2. For each of the 10 portfolios from step 1, compute and rank the year t + 1 revenue growth ((t + 1)/t) and form 10 equally sized portfolios.
Table 10 presents the results in a percentage of observations format for the transition matrix. Under the null hypotheses of no association across years in revenue growth, each cell in the 10 × 10 matrix is expected to have the same percentage of observations. Three patterns observable in Table 10 are as follows.
1. The revenue ‘momentum’ corner portfolios in Table 10 – [1,1] for successive years of lowest growth and [10,10] for successive years of highest growth – have higher than expected per-centages, as shown in Table 11.
2. Momentum in the [10,10] corner is consistent with economic factors giving current high-revenue growth companies a sustaining advantage over multiple years (e.g. due to, say, superior branding, exclusive patents, deeper customer relationships, or a sustainable low-cost advantage). Here, above-average revenue growth in year t is likely to indicate above-average revenue growth in year t + 1. Momentum in the [1,1] corner is consistent with the factors giving rise to revenue decline (e.g. poor quality problems, inexperienced manage-ment repeatedly incurring self-inflicted wounds, or aggregate market declining due to a disruptive technology) persisting over multiple years.
3. The ‘reversal’ corner portfolios in Table 10 – [1,10] and [10,1] – have above-expected per-centages for both corners of the sample one transition matrix, for neither of the reversal corners for the sample two matrix, and only the [Port 1, t; Port 10, t + 1] corner for sample three, as shown in Table 12.
Table 9. Summary data for the ratio of the 90th MKTt/REV
t percentile for negative growth group vis-a-vis
positive revenue growth group
Time period Average ratio # Years with ratio >1
Sample one: NASDAQ/NYSE/AMEX Selected SIC industries 1980–2004 2.36 22/25 All other companies 1980–2004 1.58 22/25Sample two: internet companies 1997–2004 5.26 3/8Sample three: venture-backed companies 1992–2004 31.87 13/13
NSYE: New York Stock Exchange, AMEX: American Stock Exchange, NASDAQ: National Association of Securities Dealers Automated Quotation System, SIC: standard industrial classification
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 37
4. Above-average ‘reversal’ is consistent with transitory revenue growth being a sizable component of reported revenue growth for some companies. For samples one and three, the [Port 1, t; Port 10, t + 1] combination is above average. Negative transitory revenue growth in year t will likely result in an annual revenue growth in year t + 1 that is above average(due to the ‘abnormally’ low year t reported revenue in the numerator of year t and the denominator in year t + 1).
5. Table 10 shows the row percentages for those observations with a missing revenue number in the REV
t/REV
t-1 computation. Using the portfolio cutoff points for year t + 1, we classify
those observations into one of the 10 REVt+1
/REVt portfolios.14 The percentage membership
of the 10 revenue growth portfolios in year t + 1 indicate that the year t subsample with miss-ing observations has a disproportionately higher percentage in the higher revenue growth portfolios in year t + 1. For example, the combined percentages in portfolios 9 and 10 are: sample one – selected SIC industries (52.99%); sample one – all other companies (45.32%); sample two – internet companies (56.09%); and sample three – venture-backed companies (43.81%). One explanation for sample one and two is that the newly listed IPO companies include those from sample three. Figure 4 documents the quantumly higher revenue growth rates in the private company sample.
8.2 Multi-year analysis of MKTt/REV
t portfolios with different signs of annual revenue
growthAcross each of the three samples, there is a positive revenue growth expectation. For instance, the median annual revenue growth rates are: sample one – selected SIC industries (11.4%); sample one – all other companies (8.4%); sample two – internet companies (47.1%); and sample three – venture-backed companies (56.0%). Hypothesis three is that companies with reported negative revenue growth will likely include a subset where the current reported revenue number includes a sizable negative transitory component. We now examine whether the capital market sets MKT
t/REV
t multiples recognizing this negative transitory component in an empirically
observable way. We first examine the sample with positive revenue growth observations, in part to use it as a benchmark to interpret the sample with negative revenue growth observations.
Figure 5(a) shows the median company revenue growth rates for 10 portfolios ranked on the basis of their MKT
t/REV
t values, based only on observations with positive revenue growth in year
t. These 10 portfolios comprise the top 10%, next 10%, …, bottom 10% of MKTt/REV
t observa-
tions for each company/year combination. By construction, they range from the highest to the lowest MKT
t/REV
t observations. However, this does not mechanically rank the portfolios on the
basis of revenue growth. Table 13 reports the correlation between the median MKTt/REV
t multiple
of each portfolio and its median annual revenue growth. The analysis is conducted for the annual revenue growth at year t, year t + 1, and year t + 2 for the 10 portfolios constructed in year t. For the subsample of observations with positive revenue growth in the year the MKT
t/REV
t multiple
is computed (i.e. year t), there is a significant positive correlation at the portfolio level between MKT
t/REV
t and annual revenue growth in each of years 1, 2, and 3 for samples one (both selected
SIC industries and all other companies) and for sample three. The internet sample has significant positive correlations in years 1 and 2, but not in year 3. A word of caution is appropriate in inter-preting the internet sample results. The analysis in Figure 5 is conducted in event time (rather than calendar time). The sample with the least diversification in constructing multiple sequences of calendar time to package in event time is the internet sample. The internet sample is heavily con-centrated in the 1997–2002 period and exhibits the most precipitous drop in annual revenue growth
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
38 Australian Journal of Management 36(1)Ta
ble
10.
Tran
sitio
n m
atri
x fo
r re
venu
e gr
owth
dec
ile in
yea
r t a
nd y
ear
t + 1
for
sam
ple
one
sele
cted
SIC
indu
stri
es a
nd a
ll ot
her
com
pani
es: 1
980
– 20
04. (
Port
folio
1 is
low
est
reve
nue
grow
th…
Port
folio
10
is h
ighe
st r
even
ue g
row
th.)
(a)
Sele
cted
SIC
indu
stri
es. (
b) A
ll ot
her
com
pani
es. (
c) S
ampl
e tw
o (in
tern
et c
ompa
nies
). (d
) Sa
mpl
e th
ree
(ven
ture
-bac
ked
com
pani
es)
(a)
Sam
ple
one
: Sel
ecte
d S
IC in
dust
ries
Rev
enue
gro
wth
in Y
ear
t +
1
port
1po
rt2
port
3po
rt4
port
5po
rt6
port
7po
rt8
port
9po
rt10
Tota
l
Rev
gro
wth
in y
ear
t
po
rt 1
27.4
2%14
.07%
8.35
%5.
52%
5.39
%4.
64%
4.87
%5.
69%
6.11
%17
.95%
100.
00%
po
rt 2
13.6
8%19
.40%
15.9
1%10
.16%
7.96
%7.
96%
7.73
%6.
74%
5.56
%4.
93%
100.
00%
po
rt 3
8.31
%12
.68%
17.0
2%16
.73%
12.3
6%10
.25%
7.76
%7.
03%
4.63
%3.
22%
100.
00%
po
rt 4
5.29
%8.
05%
13.0
2%21
.96%
19.0
7%12
.16%
8.28
%5.
62%
4.08
%2.
47%
100.
00%
po
rt 5
4.70
%7.
36%
9.73
%13
.71%
19.9
9%16
.80%
12.2
6%7.
50%
5.33
%2.
63%
100.
00%
po
rt 6
4.63
%7.
66%
9.92
%9.
99%
12.4
5%17
.75%
17.1
9%10
.45%
6.87
%3.
09%
100.
00%
po
rt 7
5.75
%7.
76%
7.96
%8.
05%
9.14
%12
.85%
16.4
0%16
.40%
10.6
8%5.
00%
100.
00%
po
rt 8
6.87
%7.
92%
7.56
%6.
61%
6.18
%8.
54%
12.4
2%18
.63%
18.1
7%7.
10%
100.
00%
po
rt 9
8.97
%7.
33%
6.18
%4.
31%
4.93
%6.
31%
8.45
%14
.00%
23.7
7%15
.75%
100.
00%
po
rt 1
014
.37%
7.76
%4.
37%
2.96
%2.
53%
2.76
%4.
64%
7.96
%14
.79%
37.8
7%10
0.00
%
Tota
l10
0.00
%10
0.00
%10
0.00
%10
0.00
%10
0.00
%10
0.00
%10
0.00
%10
0.00
%10
0.00
%10
0.00
%
Mis
sing
8.64
%4.
10%
3.40
%3.
31%
3.76
%5.
39%
7.37
%11
.04%
17.0
4%35
.95%
100.
00%
(b)
Sam
ple
one
: All
oth
er c
om
pani
es
Rev
enue
gro
wth
in y
ear
t +
1
port
1po
rt2
port
3po
rt4
port
5po
rt6
port
7po
rt8
port
9po
rt10
Tota
l
Rev
gro
wth
in y
ear
t
port
129
.60%
13.3
7%7.
84%
5.26
%5.
00%
4.60
%4.
92%
5.70
%7.
60%
16.1
0%10
0.00
%
port
214
.55%
19.0
6%14
.94%
10.7
8%8.
48%
7.28
%6.
80%
6.37
%6.
01%
5.72
%10
0.00
%
port
37.
96%
13.9
5%17
.38%
16.2
6%12
.12%
9.92
%7.
96%
6.01
%4.
95%
3.49
%10
0.00
%
port
45.
82%
10.2
3%13
.91%
17.5
2%16
.12%
12.2
6%9.
37%
6.64
%4.
91%
3.22
%10
0.00
%
port
54.
76%
8.37
%11
.58%
15.0
2%16
.20%
14.9
3%11
.57%
8.37
%5.
97%
3.21
%10
0.00
%
port
64.
68%
7.55
%9.
62%
10.5
2%13
.97%
16.1
9%15
.19%
10.9
6%7.
23%
4.09
%10
0.00
%
port
74.
78%
6.96
%8.
24%
8.71
%10
.85%
13.5
8%15
.77%
15.4
4%10
.41%
5.26
%10
0.00
%
port
85.
47%
7.05
%6.
38%
6.86
%7.
96%
10.2
7%13
.96%
18.0
2%16
.02%
8.03
%10
0.00
%
port
98.
39%
7.31
%5.
64%
5.60
%5.
84%
6.92
%9.
17%
14.3
9%20
.77%
15.9
7%10
0.00
%
port
10
13.9
9%6.
14%
4.46
%3.
46%
3.47
%4.
05%
5.29
%8.
09%
16.1
3%34
.93%
100.
00%
To
tal
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
M
issi
ng8.
36%
5.35
%6.
02%
5.45
%5.
55%
6.28
%7.
58%
10.0
8%14
.23%
31.0
9%10
0.00
%
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 39
(c)
Sam
ple
two
: Int
erne
t co
mpa
nies
Rev
enue
gro
wth
in y
ear
t +
1
port
1po
rt2
port
3po
rt4
port
5po
rt6
port
7po
rt8
port
9po
rt10
Tota
l
Rev
gro
wth
in y
ear
t
port
126
.67%
20.0
0%13
.94%
7.27
%9.
09%
6.06
%4.
24%
2.42
%4.
24%
6.06
%10
0.00
%
port
211
.45%
19.8
8%21
.69%
15.0
6%15
.06%
9.64
%4.
22%
1.20
%1.
20%
0.60
%10
0.00
%
port
34.
82%
10.8
4%20
.48%
28.3
1%18
.07%
11.4
5%3.
01%
1.20
%1.
20%
0.60
%10
0.00
%
port
46.
06%
7.27
%10
.91%
18.7
9%18
.18%
16.3
6%13
.94%
3.64
%1.
82%
3.03
%10
0.00
%
port
57.
83%
6.63
%6.
02%
9.04
%16
.27%
17.4
7%19
.88%
7.83
%6.
02%
3.01
%10
0.00
%
port
67.
23%
10.2
4%6.
02%
7.83
%8.
43%
13.2
5%21
.69%
12.6
5%7.
23%
5.42
%10
0.00
%
port
713
.33%
9.70
%8.
48%
4.85
%4.
85%
6.67
%10
.91%
19.3
9%16
.97%
4.85
%10
0.00
%
port
89.
64%
6.02
%6.
02%
3.01
%4.
22%
6.02
%12
.65%
25.3
0%16
.87%
10.2
4%10
0.00
%
port
97.
23%
7.23
%4.
82%
4.22
%3.
01%
7.23
%5.
42%
15.6
6%21
.08%
24.1
0%10
0.00
%
port
10
5.45
%2.
42%
1.82
%1.
21%
3.03
%6.
06%
3.64
%10
.91%
23.6
4%41
.82%
100.
00%
To
tal
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
M
issi
ng2.
59%
1.20
%0.
80%
2.20
%5.
59%
7.19
%11
.98%
12.3
8%16
.77%
39.3
2%10
0.00
%
(d)
Sam
ple
thre
e: V
C p
riva
te c
om
pani
es
Rev
enue
gro
wth
in y
ear
t +
1
port
1po
rt2
port
3po
rt4
port
5po
rt6
port
7po
rt8
port
9po
rt10
Tota
l
Rev
gro
wth
in y
ear
t
port
121
.25%
12.8
9%9.
76%
5.92
%6.
62%
6.62
%6.
62%
9.06
%7.
32%
13.9
4%10
0.00
%
port
213
.89%
24.6
5%20
.49%
12.8
5%7.
99%
4.86
%5.
21%
3.82
%2.
78%
3.47
%10
0.00
%
port
310
.76%
14.2
4%21
.53%
16.6
7%12
.85%
6.60
%5.
56%
4.51
%3.
13%
4.17
%10
0.00
%
port
410
.42%
15.9
7%13
.19%
19.4
4%17
.01%
10.4
2%6.
25%
2.78
%3.
47%
1.04
%10
0.00
%
port
58.
68%
6.94
%12
.50%
12.1
5%16
.32%
12.8
5%12
.85%
7.29
%6.
25%
4.17
%10
0.00
%
port
69.
38%
9.72
%5.
56%
12.1
5%12
.85%
16.3
2%14
.58%
10.0
7%5.
90%
3.47
%10
0.00
%
port
77.
99%
5.90
%6.
94%
7.29
%11
.46%
17.0
1%12
.50%
14.5
8%10
.42%
5.90
%10
0.00
%
port
86.
55%
4.14
%2.
41%
4.48
%6.
90%
14.4
8%15
.17%
20.6
9%13
.45%
11.7
2%10
0.00
%
port
95.
59%
4.55
%4.
90%
6.29
%4.
55%
6.29
%16
.43%
16.7
8%19
.23%
15.3
8%10
0.00
%
port
10
5.23
%1.
05%
2.79
%2.
79%
3.48
%4.
53%
4.88
%10
.45%
28.2
2%36
.59%
100.
00%
To
tal
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
100.
00%
M
issi
ng10
.51%
7.10
%5.
97%
5.38
%6.
22%
6.96
%6.
86%
7.20
%12
.78%
31.0
3%10
0.00
%
Tabl
e 10
. (C
ontin
ued)
SIC
: sta
ndar
d in
dust
rial
cla
ssifi
catio
n
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
40 Australian Journal of Management 36(1)
in 2000 and beyond (see Figure 4). In contrast, samples one and three have longer time sequences to effect the time-period diversification.
The general finding from Figure 5(a) is that the capital market sets higher MKTt/REV
t multiples
for companies with positive revenue growth not only in year t, but also with the realized growth rates in years t + 1 and t + 2. For the observations with positive annual revenue growth, there is support for both the sign and magnitude of past revenue growth, as well as subsequent realized revenue growth being positively associated with MKT
t/REV
t multiples.
Figure 5(b) and Table 10 both report results for those observations with negative revenue growth. There is a key dramatic difference from the positive revenue growth observations in Figure 5(a). In the year the 10 MKT
t/REV
t portfolios are formed (i.e. year t), there is a high negative cor-
relation between the median MKTt/REV
t in year t of each portfolio and the magnitude of the rev-
enue growth in year t. This significant negative correlation supports the hypothesis that reported revenue in year t includes a negative transitory revenue component that the capital market takes into account. The more negative the transitory revenue component, the lower the denominator of the MKT
t/REV
t computation. Hypothesis three can be empirically probed by observing the real-
ized annual revenue growth rates in years t + 1 and t + 2 for the 10 portfolios in Figure 5(b). The revenue growth rates in years t + 1 and t + 2 are positively correlated with the MKT
t/REV
t multi-
ples. This flipping of the sign and magnitude of the portfolio revenue growth rates between year t and year t + 1 in Figure 5(b) is most marked for samples one and three. Our results are consistent with hypothesis three – the capital market sets MKT
t/REV
t multiples such that the magnitude of
transitory revenue is recognized; higher MKTt/REV
t multiples are assigned to those negative year
t revenue growth companies with higher realized (and presumably expected) annual revenue growth in years t + 1 and t + 2.
Table 11. The revenue ‘momentum’ corner portfolios in Table 10
Portfolio 1, t/Portfolio 1, t + 1
Portfolio 10, t/Portfolio 10, t + 1
Sample one: NASDAQ/NYSE/AMEX Selected SIC industries 27.42% 37.87% All other companies 29.60% 34.93%Sample two: internet companies 26.67% 41.82%Sample three: venture-backed companies 21.15% 36.59%
NSYE: New York Stock Exchange, AMEX: American Stock Exchange, NASDAQ: National Association of Securities Dealers Automated Quotation System, SIC: standard industrial classification
Table 12. Percentages for the ‘reversal’ corner portfolios in Table 10
Portfolio 1, t/Portfolio 10, t/
Portfolio 10, t + 1Portfolio 1, t + 1
Sample one: NASDAQ/NYSE/AMEX Selected SIC industries 17.95% 14.37% All other companies 16.10% 13.99%Sample two: internet companies 6.06% 5.45%Sample three: venture-backed companies 13.94% 5.23%
NSYE: New York Stock Exchange, AMEX: American Stock Exchange, NASDAQ: National Association of Securities Dealers Automated Quotation System, SIC: standard industrial classification
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 41
Sample OneSelected SIC Industries (Public)
Sample OneAll Other Companies (Public)
Yea
r t
Yea
r t+
1Y
ear
t+2
Sample Two – Internet (Public) Sample Three –VC Backed (Private)
Yea
r t
Yea
r t+
1Y
ear
t+2
(a)
0%
10%
20%
30%
40%
50%
60%
70%
80%
10 9 8 7 6 5 4 3 2 1
Positive Revenue Growth
0%
10%
20%
30%
40%
50%
60%
70%
80%
10 9 8 7 6 5 4 3 2 1
Positive Revenue Growth
0%
10%
20%
30%
40%
50%
60%
10 9 8 7 6 5 4 3 2 1
Positive Revenue Growth
0%
10%
20%
30%
40%
50%
60%
10 9 8 7 6 5 4 3 2 1
Positive Revenue Growth
0%
5%
10%
15%
20%
25%
30%
10 9 8 7 6 5 4 3 2 1
Positive Revenue Growth
0%
5%
10%
15%
20%
25%
30%
10 9 8 7 6 5 4 3 2 1
Positive Revenue Growth
0%
20%
40%
60%
80%
100%
120%
140%
160%
180%
10 9 8 7 6 5 4 3 2 1
Revenue Growth Portfolio
0%
1000%
2000%
3000%
4000%
5000%
6000%
7000%
8000%
9000%
10000%
10 9 8 7 6 5 4 3 2 1
Revenue Growth Portfolio
0%
20%
40%
60%
80%
100%
120%
140%
160%
10 9 8 7 6 5 4 3 2 1
Revenue Growth Portfolio
0%
20%
40%
60%
80%
100%
120%
140%
160%
10 9 8 7 6 5 4 3 2 1
Revenue Growth Portfolio
0%
20%
40%
60%
80%
100%
120%
10 9 8 7 6 5 4 3 2 1
Revenue Growth Portfolio
0%
20%
40%
60%
80%
100%
120%
10 9 8 7 6 5 4 3 2 1
Revenue Growth Portfoli0
(Continued)
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
42 Australian Journal of Management 36(1)
Sample OneSelected SIC Industries (Public)
Sample OneAll Other Companies (Public)
Yea
r t
Yea
r t+
1Y
ear
t+2
Sample Two – Internet (Public) Sample Three – VC Backed (Private)
Yea
r t
Yea
r t+
1Y
ear
t+2
-60%
-50%
-40%
-30%
-20%
-10%
0%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
-60%
-50%
-40%
-30%
-20%
-10%
0%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
-20%
-10%
0%
10%
20%
30%
40%
50%
60%
70%
80%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
-20%
-10%
0%
10%
20%
30%
40%
50%
60%
70%
80%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
-10%
-5%
0%
5%
10%
15%
20%
25%
30%
10 9 8 7 6 5 4 3 2 1-10%
-5%
0%
5%
10%
15%
20%
25%
30%
10 9 8 7 6 5 4 3 2 1
-80%
-70%
-60%
-50%
-40%
-30%
-20%
-10%
0%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
-80%
-70%
-60%
-50%
-40%
-30%
-20%
-10%
0%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
-250%
-200%
-150%
-100%
-50%
0%
50%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
-50%
0%
50%
100%
150%
200%
250%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
-20%
0%
20%
40%
60%
80%
100%
120%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
-20%
0%
20%
40%
60%
80%
100%
120%
10 9 8 7 6 5 4 3 2 1
Negative Revenue Growth
(b)
Figure 5. (a) Median annual portfolio revenue growth in years t, t + 1, and t + 2 for portfolios formed using MKT
t/REV
t rankings: 10 portfolios with positive revenue growth in year t. (b) Median annual portfolio
revenue growth in years t, t + 1, and t + 2 for portfolios formed using MKTt/REV
t rankings: 10 portfolios
with negative revenue growth in year t. SIC: standard industrial classification, VC: venture capital.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 43
Table 13. Correlation for 10 portfolios ranked on MKTt/REV
t between (a) median MKT
t/REV
t and median
annual revenue growth, and (b) annual revenue growth across time periods
(a)
Sample one Sample two Sample three
Selected SIC industries All other companies Internet companies VC-backed companies
Positive Revenue Growth Portfolios MKT
t/REVt & REV growth
t/t–1
Pearson 0.97 0.99 0.95 0.83 Spearman 1.00 0.98 0.60 0.92 MKT
t/REV
t & REV growth
t+1/t
Pearson 0.93 0.97 0.98 1.00 Spearman 1.00 1.00 0.96 1.00 MKT
t/REV
t & REV growth
t+2/t+1
Pearson 0.86 0.92 -0.04 0.99 Spearman 1.00 1.00 0.17 0.98Negative revenue growth portfolios MKT
t/REV
t & REV growth
t/t–1
Pearson –0.98 -0.99 -0.93 -0.92 Spearman –0.80 -0.51 0.01 -0.83 MKT
t/REV
t & REV growth
t+1/t
Pearson 0.97 0.89 0.87 0.79 Spearman 1.00 0.99 0.78 0.90 MKT
t/REV
t & REV growth
t+2/t+1
Pearson 0.91 0.72 0.59 0.70 Spearman 0.99 0.96 0.45 0.89
(b)
Sample one Sample two Sample three
Selected SIC industries All other companies Internet companies VC-backed companies
Positive revenue growth portfolios Year t & year t + 1 Pearson 0.99 0.95 0.91 0.83 Spearman 1.00 0.99 0.71 0.92 Year t & year t + 2 Pearson 0.94 0.89 –0.03 0.86 Spearman 0.99 0.99 0.17 0.96 Year t + 1 & year t + 2 Pearson 0.98 0.98 0.06 0.99 Spearman 1.00 1.00 0.31 0.98Negative revenue growth portfolios Year t & year t + 1 Pearson -0.98 -0.82 -0.76 -0.90 Spearman -0.80 -0.46 -0.01 -0.89 Year t & year t + 2 Pearson -0.95 -0.62 -0.63 -0.84 Spearman -0.78 -0.34 -0.42 -0.74 Year t + 1 & year t + 2 Pearson 0.98 0.95 0.45 0.90 Spearman 0.99 0.98 0.37 0.80
VC: venture capital
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
44 Australian Journal of Management 36(1)
Table 13 shows results for: (a) the correlations between median MKTt/REV
t and median annual
revenue growth in years t, t + 1, and t + 2 (Table 13(a)), and (b) the correlations between median annual revenue growth of the 10 MKT
t/REV
t portfolios for pairs of years t, t + 1, and t + 2 (in Table
13(b)). The correlations in Table 13(b) highlight the revenue momentum factor for the positive reve-nue growth subsample. It is the strongest in samples one and three. The internet companies in sample two exhibit significant evidence of revenue momentum between years t and t + 1, but none between years t and t + 2 or between years t + 1 and t + 2. The results for the negative revenue growth portfolios in Table 13(b) highlight the reversal pattern associated with the negative transitory revenue compo-nent in year t. Note also the positive correlations between the median annual revenue growth rates across portfolios between years t + 1 and t + 2 (with the internet sample having the lowest positive correlation). Portfolios formed on MKT
t/REV
t in year t for a sample with negative revenue growth
have positive Pearson correlations between median annual revenue growth rates in years t + 1 and years t + 2 of 0.98 (sample one – selected SIC), 0.95 (sample one – all other), 0.45 (sample two – internet), and 0.90 (sample three – venture-backed). This is strong support for our negative transitory revenue hypothesis for the MKT
t/REV
t multiples of negative revenue growth companies in year t.
9. Multivariate analysisThis section incorporates our prior analysis into a multivariate framework. Table 14 shows the regres-sion structure that further probes the three hypotheses for our three samples. In Table 15 we present results for rank regressions due to the extreme observations found in each of the three samples.
Table 14. Regression structure
Dependent variable MKTt/REV
tIndependent variables Company size log(REV
t)
Current revenue growth [(REVt/REV
t–1) – 1]
Future revenue growth [(REVt+1
/REVt) – 1]
Profitability Income/revenue if >0; Neg. income/revenue if ≤0 Leverage Long-term debt/Total assets 1998–2000 time period Indicator = 1 if 1998–2000
Table 15 reports results for the regression with combinations of independent variables. Regression #1 reports results that pertain to hypothesis one, which predicts a negative relationship between MKT
t/
REVt and company size. Across each of the samples for regression #1, this hypothesis is strongly sup-
ported. Hypothesis two – companies with higher positive current revenue growth will have higher MKT
t/REV
t – is consistently supported by the regression #2 results. Regressions #3–#5 combine
variables related to company size, current revenue growth, and future revenue growth (defined as the realized revenue growth in year t + 1/year t). The future revenue growth variable is added in regres-sions #3b, #4, and #5 to recognize that the capital market uses a broad-based information set when forecasting future revenue growth. Both the current revenue growth and the future revenue growth measures are expected to have measurement error regarding capturing the expected revenue growth implicit in MKT
t/REV
t. The current revenue growth variable (REV
t/REV
t–1) will incorporate factors
that affect year t but are not expected to affect subsequent years. The future revenue growth variable (REV
t+1/REV
t–1) will incorporate factors affecting year t + 1 that were unanticipated at the end of year
t. Both measures also are scalars (representing one year of revenue growth) whereas the capital market likely is impounding a vector (representing revenue growth over a sequence of future years). The regression results in regressions #1–#5 consistently support hypotheses one and two.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 45
Tabl
e 15
. R
egre
ssio
n an
alys
is (
rank
reg
ress
ion)
usi
ng (
a) M
KT
t/REV
t as
depe
nden
t va
riab
le, a
nd (
b) r
even
ue g
row
th, c
ompa
ny s
ize,
pro
fitab
ility
, and
oth
er
vari
able
s
Sam
ple
one
Sam
ple
two
Sam
ple
Thr
ee
Sele
cted
SIC
All
othe
rIn
tern
etV
C-b
acke
d
Coe
f.t–
stat
C
oef.
t–st
at
Coe
f.t–
stat
Coe
f.t–
stat
Reg
ress
ion
#1In
terc
ept
1953
820
2.73
8578
943
1.18
1085
38.8
211
0897
.45
log(
REV
)-0
.29
-52.
20-0
.31
-117
.19
-0.2
1-1
1.07
-0.8
3-5
0.87
Adj
. R2
8.2%
9.5%
7.1%
68.1
%R
egre
ssio
n #2
Inte
rcep
t10
508
104.
0452
538
246.
4340
021
.31
385
16.8
8C
urre
nt r
ev g
row
th0.
2636
.82
0.17
49.2
50.
3919
.00
0.32
6.14
Adj
. R2
5.2%
2.2%
20.8
%4.
7%R
egre
ssio
n #3
Inte
rcep
t14
875
114.
6372
816
261.
3172
423
.68
1022
43.1
8lo
g(R
EV)
-0.2
6-4
5.07
-0.2
7-9
7.74
-0.1
9-1
1.28
-0.7
8-3
3.90
Cur
rent
rev
gro
wth
0.29
43.0
90.
1755
.05
0.39
20.2
30.
154.
64A
dj. R
212
.0%
9.6%
25.8
%62
.1%
Reg
ress
ion
#3b
Inte
rcep
t14
750
118.
4769
869
266.
4562
420
.32
946
47.0
0lo
g(R
EV)
-0.2
7-5
1.38
-0.2
9-1
15.2
6-0
.13
-8.2
2-0
.73
-40.
11Fu
ture
rev
gro
wth
0.30
56.8
00.
2389
.75
0.39
24.0
10.
189.
61A
dj. R-s
q17
.1%
14.7
%31
.7%
70.4
%R
egre
ssio
n #4
Inte
rcep
t78
7666
.89
4358
317
5.96
261
14.0
5923
29.
63C
urre
nt r
ev g
row
th0.
1927
.37
0.10
29.1
50.
2210
.88
0.11
2.19
Futu
re r
ev g
row
th0.
2540
.18
0.21
67.3
20.
3217
.89
0.45
12.4
6A
dj. R
211
.1%
6.0%
35.7
%20
.9%
(Con
tinue
d)
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
46 Australian Journal of Management 36(1)
Sam
ple
one
Sam
ple
two
Sam
ple
Thr
ee
Sele
cted
SIC
All
othe
rIn
tern
etV
C-b
acke
d
Coe
f.t–
stat
C
oef.
t–st
at
Coe
f.t–
stat
Coe
f.t–
stat
Reg
ress
ion
#5In
terc
ept
1139
481
.661
103
203.
6944
515
.282
936
32.4
0lo
g(R
EV)
-0.2
5-4
3.27
-0.2
7-9
5.92
-0.1
3-8
.07
-0.7
3-2
9.62
Cur
rent
rev
gro
wth
0.22
32.6
80.
1133
.68
0.24
11.7
60.
103.
02Fu
ture
rev
gro
wth
0.24
39.8
90.
2066
.61
0.29
16.3
50.
135.
07A
dj. R
217
.3%
13.3
%38
.6%
63.3
%R
egre
ssio
n #6
Inte
rcep
t-7
539
-19.
27-5
1279
-97.
2810
6610
8.94
POSI
NC
/REV
0.57
47.9
40.
8424
5.32
-0.3
30.
08N
EG IN
C/R
EV0.
9069
.37
0.90
199.
91-0
.14
0.08
Adj
. R2
15.7
%30
.2%
4.3%
Reg
ress
ion
#7In
terc
ept
-779
3-2
5.15
-246
75-4
6.36
138
1.31
982.
8241
.76
log(
REV
)-0
.21
-36.
27-0
.22
-90.
81-0
.17
-4.3
5-0
.771
5-3
3.98
15C
urre
nt r
ev g
row
th0.
1529
.38
0.04
19.1
90.
020.
840.
1346
24.
3442
3Fu
ture
rev
gro
wth
0.25
49.6
10.
1879
.31
0.45
19.4
42.
7949
42.
5140
3PO
SIN
C/R
EV0.
6556
.61
0.76
233.
640.
202.
01N
EG IN
C/R
EV0.
9073
.12
0.75
163.
210.
303.
43Le
vera
ge-0
.16
-29.
39-0
.10
-45.
38-0
.14
-5.8
919
98-2
000
dum
my
1993
19.5
441
3117
.87
190
10.4
818
1.21
410
.379
Adj
. R2
36.9
%43
.6%
40.1
%67
.2%
NSY
E: N
ew Y
ork
Stoc
k Ex
chan
ge, A
MEX
: Am
eric
an S
tock
Exc
hang
e, N
ASD
AQ
: Nat
iona
l Ass
ocia
tion
of S
ecur
ities
Dea
lers
Aut
omat
ed Q
uota
tion
Syst
em, S
IC: s
tand
ard
indu
stri
al c
lass
ifica
tion,
VC
: ven
ture
cap
ital
Tabl
e 15
. (C
ontin
ued)
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 47
Existing textbooks emphasize the importance of profitability in capital market valuation – see, for example, Palepu and Healy (2007) and Penman (2009). Regression #6 reports regression results for MKT
t/REV
t as the dependent variable and two related profitability independent variables:
• POSINC/REV (net income/revenues) – if net income is positive or zero;• NEGINC/REV (negative net income/revenues).
We include these two separate profitability variables due to prior research documenting ‘anom-alous’ results for the loss company observations.15 Results are reported for the two publicly traded company samples; the VentureSource database has limited net income observations for the pri-vately held sample three companies. For both sample one subsamples the coefficients on the two profitability variables are positive and statistically significant – the more profitable the company, the higher the MKT
t/REV
t multiple. However, the internet companies show insignificant coeffi-
cients on the two profitability variables in regression #6.Regression #7 pools the independent variables in regressions #1–#6. We also include two addi-
tional variables:
• financial leverage (long-term debt to total assets);• 1998–2000 time period intercept dummy variable.
The rank regressions in regression #7 provide strong support for hypotheses one and two in a multivariate context.
Sample two (internet companies) alone has several changes in the sign or significance of indi-vidual independent variables for regression #7 vis-à-vis regressions #1–#6. In regression #7, the coefficient on the future revenue growth for sample two is positive and significant, whereas the coefficient on the current revenue growth is positive but insignificant; both growth variables are positive and significant in regressions #4 and #5. Regression #7 reports significant positive coef-ficients on the two profitability variables for sample two, whereas they were insignificantly nega-tive in regression #6.
A comparison of the R2 values across selected regressions provides additional insight into the differing significance of the company size, revenue growth, and profitability variables. We succes-sively use the adjusted R2 values in regressions #1 (company size), #4 (revenue growth), and #6 (profitability) as the numerator and the adjusted R2 value in regression #7 as the denominator16 (see Table 16).
Table 16.
Sample one Sample two internet companies
Sample three VC-backed private companiesSelected SIC
industriesAll other companies
Company size (#1/#7)
8.2%36.9%
= 22.2%9.5%43.6%
= 21.8%7.1%40.1%
= 17.7%68.1%67.2%
= 101.3%
Revenue growth(#4/#7)
11.1%36.9%
= 30.1%6.0%43.6%
= 13.8%35.7%40.1%
= 89.0%20.9%67.2%
= 31.1%
Profitability(#6/#7)
15.7%36.9%
= 42.5%30.2%43.6%
= 69.3%4.3%40.1%
= 10.7%
VC: venture capital
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
48 Australian Journal of Management 36(1)
For the two sample one subgroups of companies, profitability variables contribute the most explanatory power. Revenue growth is more important for the selected SIC industries than for the all other companies; the R2 values for revenue growth variables in regression #4 are 30.1% of the regression #7 R2 value for the selected SIC industries subsample, but only 13.8% for the all other companies subsample. Revenue growth is the dominant contributor for sample two (internet com-panies), while company size is the dominant contributor for sample three (venture-backed private companies).
Table 17 reports for sample one rank regressions for the (a) positive revenue growth, (b) nega-tive revenue growth, and (c) missing revenue growth observation subsamples in panels (a)–(c), respectively. The positive revenue growth subsample is the largest subsample and has results simi-lar to regression #7 in Table 15. The negative revenue growth subsample has a sign reversal on the two revenue growth independent variables – the coefficient on current revenue growth is signifi-cantly negative, while the coefficient on future revenue growth is significantly positive. These results support hypothesis three and are consistent with the ‘flipping’ of the portfolio revenue growth rates in year t and year t + 1 in Figure 5(b). The missing observations subsample has (by definition) no current revenue growth independent variable; the coefficient on the future revenue growth variable is significantly positive.
The multivariate results in Tables 15 and 17 reinforce the findings reported in prior sections. Revenue growth and company size are key determinants of MKT
t/REV
t multiples. Moreover, the
capital market is able to distinguish (in a probabilistic sense) cases where both current and future revenue growth are positively correlated from cases where they are negatively correlated.
10. Overview and extensions of researchOur findings have implications for research in at least four areas – capital market valuation, ‘man-agement’ of reported financial numbers, analysis of the 1998–2000 capital market ‘bubble era’, and early-stage company valuation and growth.
10.1 Capital market valuation researchPrior capital market research has recognized that deleting observations with negative net income or negative book value results in samples under-represented in:
a) high-technology companies; andb) younger early-stage companies.
The samples examined in this paper include a broad cross-section of (a) and (b) companies, as well as companies from other industries and in their post-early stages. We document several marked features of (a) and (b) companies. Such companies typically have higher revenue growth rates, higher variability in revenue growth rates, and a higher negative income likelihood. The strong support for hypothesis one (negative association between MKT
t/REV
t and company size)
and hypothesis two (positive association between MKTt/REV
t and recent positive revenue
growth) highlights the importance of recognizing company size and revenue growth variables in company valuation research. Note, moreover, that these variables are also highly statistically significant in a broad sample of non-high-technology publicly traded companies (sample one – all other companies).
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 49
Table 17. Sample one regression analysis (rank regression) using MKTt/REVt as dependent variables for three subsamples based on sign of current revenue growth: (a) positive; (b) negative; (c): missing observations)
Panel A: Positive current revenue growth
Selected SIC All other
Coef. t–stat Coef. t–stat
Intercept -5830.62 -29.74 -20794.45 -54.55log(REV) -0.06 -8.43 -0.13 -46.55Current rev growth 0.22 33.17 0.10 34.98Future rev growth 0.18 27.40 0.14 49.79POSINC/REV 0.66 58.83 0.77 225.51NEG INC/REV 0.89 65.96 0.76 135.24Leverage -0.15 -22.73 -0.08 -29.731998–2000 period 1075.88 14.10 1937.86 11.55
R2 41.68% 49.37%nobs 15,131 73,574
Panel B: Negative current revenue growth
Selected SIC All other
Coef. t-stat Coef. t-stat
Intercept 1198.46 5.32 -1170.76 -3.65log(REV) -0.28 -17.90 -0.30 -54.19Current rev growth -0.09 -6.82 -0.11 -19.60Future rev growth 0.20 18.35 0.16 33.26POSINC/REV 0.34 8.58 0.80 81.67NEG INC/REV 0.56 15.22 0.64 56.65Leverage -0.13 -11.23 -0.12 -22.851998-2000 period 252.60 4.53 695.91 5.52R2 26.62% 32.89%nobs 6353 27,929
Panel C: Missing current revenue growth
Selected SIC All other
Coef. t-stat Coef. t-stat
Intercept -7.75 -0.07 -1007.32 -5.39log(REV) -0.26 -20.02 -0.36 -55.49Current rev growthFuture rev growth 0.25 23.62 0.24 42.57POSINC/REV 0.39 15.26 0.57 66.29NEG INC/REV 0.79 28.56 0.78 64.24Leverage -0.20 -18.66 -0.11 -19.511998-2000 period 277.61 8.08 1145.19 12.80R2 56.56% 53.69%nobs 4064 15681
SIC: standard industrial classification
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
50 Australian Journal of Management 36(1)
Our analysis is of increased interest in the post-‘internet bubble’ period given the continued high profile valuation of select individual early-stage companies, for example, Facebook (2004/2005 startup), Zynga (2007 startup), and Groupon (2008 startup).
10.2 Management of reported financial numbers researchThe earnings management literature hypothesizes that management proactively uses accounting rules (such as revenue or expense recognition) to change reported accounting numbers. One expla-nation for a transitory component in reported numbers is management accounting ‘adjustments’ for reasons related to misrepresenting underlying economics, for example, changing software revenue recognition assumptions to maximize a compensation bonus or an earn-out provision. However, other explanations (of a non-management manipulation kind) for the transitory component exist – for example, a negative transitory revenue component could be due to a delay in a recurring com-puter hardware sales shipment cycle due to an unexpected shortage of component parts. In some cases, a negative transitory revenue component in year t may well result in a positive transitory revenue component in year t + 1. Research on ways to better identify contexts where transitory revenue is sizable (and their possible explanations) can assist the probing of earnings ‘manage-ment’ hypotheses. The research in this paper used the joint combination of (i) a high MKT
t/REV
t multiple, and (ii) a negative current revenue growth to identify negative transitory revenue con-texts. Alternative combinations of MKT
t/REV
t levels and revenue growth rates could potentially
be used to identify positive transitory revenue contexts. Note that a key assumption here is that the capital market is able to identify (in a relatively high probabilistic sense) this transitory component when setting MKT
t/REV
t multiples.
10.3 Analysis of 1998–2000 capital market ‘bubble era’The working assumption in many descriptions of the 1998–2000 capital market period is that com-pany fundamentals did not explain the historically high levels of market multiples. Our research looks at both the pre and post behavior of revenue growth and documents their dramatic decline in 2000/2001 vis-à-vis 1998–2000. We document that industries with the largest increases/decreases in market capitalization in the 1998–2001 period had characteristics that identified them as rela-tively high risk for an extended period prior to 1998 for example, higher beta, higher standard deviation of security returns, and higher percentage of negative net income. Our results are consis-tent with capital markets behaving more rationally along the predictions of capital market theory than many observers are willing to recognize.
Our results also highlight that the capital market 1998–2000 surge and subsequent decline in valuations was not restricted to the so-called internet companies that have been the focus of many research studies. While internet companies exhibited very high positive (negative) rates of returns in the 1998–2000 (2000-–2001) period, the dollar amount of their aggregate market capitalization increase/decrease is dwarfed by the aggregate increase/decrease in market capitalization of our selected SIC industries group, which comprise the broader computer software/hardware/telecom-munications/biotech-pharmaceuticals sectors. Aggregate market capitalizations of our selected SIC industries group and the internet companies are shown in Table 18 (from Table 2 – $ billions).
Research probing the 1998–2001 capital market era needs to examine a broader set of compa-nies than just those labeled as ‘internet stocks’. Most internet companies were listed on the NASDAQ. Many of the companies in our selected SIC industries group were listed on the NYSE. The above market capitalization changes highlight that the so-called ‘internet bubble’ is an overly
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 51
narrow description. The increases/decreases in market capitalization in the 1998–2001 period occurred for many non-internet companies and for many non-NASDAQ companies.
10.4 Early-stage company researchOur research documents that both company size and revenue growth are statistically significant independent variables in MKT
t/REV
t models in both public and private markets. We also find the
dramatically higher revenue growth rates for our venture-backed private companies sample (sam-ple three) vis-à-vis the revenue growth rates for the publicly traded company samples (samples one and two) – see Figure 4. Not all early-stage companies are VC backed. Venture capitalists tradition-ally restrict themselves to companies that both address large markets and have the capacity to grow rapidly. The revenue growth rates we report in Figure 4 and Table 7 are consistent with multiple early-stage private companies achieving revenue growth rates not accomplished by many publicly traded companies. Subsequent research could examine the growth rates on non-venture financed private companies. Such analysis could be part of a broader focus on factors that explain the level and change in revenue growth rates as a company evolves over time. Our research highlights that the existence of VC financing likely is a facilitator of high-revenue growth. Other potential vari-ables that could be examined include the product/service markets targeted by a company and the experience/aspirations of the management team.
The growing analysis of early-stage companies offers many avenues to analyse how key factors affect company valuation and growth- for example, the type of private financing, the role of public listing, the chosen business model and its ability to scale, and the adoption of management control systems. The dramatic time-series and cross-sectional variations in both the capital market vari-ables and the fundamental accounting variables across both the public and private company sam-ples examined in this paper highlight the fruitful opportunities for subsequent research.
Early-stage companies are also of much interest due to many having a relatively high intangible asset composition vis-à-vis tangible assets. VC funding often occurs because companies have neg-ative net cash flows due to their making ‘investments’ in intangibles, such as technologies, discov-eries, etc., or the buildup of a sales/marketing capability. Revenues are an important signal that such ‘investments’ are being commercially valued by customers or partners. Rapid growth in rev-enues potentially is a lead indicator of subsequent growth in profitability. An important research area is understanding the strength, time horizon and determinants of such lead/lag relationships between revenue growth and profitability growth.
Table 18. Aggregate market capitalizations of our selected SIC industries group and the internet companies
Selected SIC industries
Internet companies
Selected SIC industries excluding overlapping internet companies
Aggregate market capitalizations 1 January 1998 $2489 $80 $2412 10 March 2000 9262 1630 7762 31 December 2001 4593 279 4337Aggregate changes 1 January 1998–10 March 2000 $6773 $1550 $5350 10 March 2000–31 December 2001 4669 1351 3425
SIC: standard industrial classification
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
52 Australian Journal of Management 36(1)
AcknowledgementsThe comments of Ray Ball, Bill Beaver, Philip Brown, Ron Kasznik, Charles Lee and participants at work-shops at Stanford University and the University of New South Wales are greatly appreciated. We thank Dow Jones VentureSource for access to their proprietary data base.
FundingChristopher Armstrong is grateful for financial support from the Dorinda and Mark Winkelman Distinguished Scholar Award. George Foster thanks the support of the Center for Entrepreneurial Studies at the Graduate School of Business, Stanford University.
Notes 1 See, for instance, O’Hara and Lowry (2009). 2 The Institutional Brokers Estimate System (IBES) started reporting security analyst revenue forecasts
approximately 20 years after it started reporting EPS forecasts. It started reporting one-year ahead and two-year ahead revenue (EPS) forecasts in 1996 (1976). It started reporting long-term revenue (EPS) growth forecasts in 1999 (1981). By 2004, 85.18%/83.68% of companies in the IBES sample of compa-nies had one-year ahead/two-year ahead security analyst revenue forecasts; the comparable percentages for one-year ahead/two-year ahead EPS forecasts were 90.87%/89.50%. Ertimur et al. (2011) probe explanations for the issuance of revenue and expense forecasts by security analysts.
3 An analysis of Compustat companies listed on the NYSE, AMEX, or NASDAQ finds only 15 negative annual revenue observations over the 1980–2004 period. We examined the respective 10Ks of each of the 15 negative revenue observations – 3 were errors in Compustat.
4 The analysis used to identify the two sample one sub-samples is conducted at the three-digit SIC industry level and not the company level. Our industry market capitalization amounts are based on all companies listed at the chosen date of interest. As new companies list or existing companies delist, there will be a changing number of companies underlying the aggregate industry market capitalizations at different dates.
5 10 March 2000 is the date when the NASDAQ reached its peak value (up through 31 December 2004). 6 There is no universally agreed upon definition of an internet company. The accounting literature has
relied heavily on the InternetStockList™ (ISL) reported on http://www.internet.com. The ISL was billed by http://www.internet.com as a complete list of all publicly traded internet stocks. An internet stock was operationally defined as a stock that existed because of the internet – that is, had there been no internet, the stock would not be in existence. Papers that use the ISL include Bartov et al. (2002), Bauman et al. (2010), Davis (2002), Demers and Lev (2000), Demers and Lewellen (2003), Hand (2001), Keating et al. (2003), and Trueman et al. (2000, 2001). Depending on whether the authors were targeting all internet firms or just a subset (e.g. only B2C firms), sample sizes these papers analysed range between 24 and 261. The finance literature approach appears to be to identify which new IPOs have an internet connec-tion. The Ritter database has this approach, as does Ljungqvist and Wilhem (2003), Ofek and Richardson (2002, 2003), and Schultz and Zaman (2001). The samples in these papers ranged from 393 to 538.
7 The internet list includes 4.1% that went IPO prior to 1996. One notable company included in the list is Cisco Systems (IPO in 1990).
8 One caveat on the sizably higher levels of the sample three MKTt/REV
t multiples is that VentureOne
uses as one of its information sources the S-1 filings of companies attempting to have an IPO. This may bias the private company sample to the more successful venture-backed private companies. However, many non-IPO exits for such private companies are in trade sales (acquisitions), for which market com-petition can be intense. See Appendix A of Armstrong et al. (2006) for discussion and evidence of pos-sible biases from a sample that is exclusively comprised of venture-backed private companies that filed for an IPO.
9 Ball and Foster (1982) and Foster (1986: 238–245) provide reviews of this early literature.10 Samples two and three have too limited a time series for individual companies to compute comparable
autocorrelations.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 53
11 The higher risk selected SIC industries subsample had dramatically higher positive security returns in the 1998–2000 period. Capital market theory would predict that these companies would also have dramati-cally more negative security returns in the event of a market decline. This is exactly what happened in December 2001. See Table 2.
12 For companies in sample three, there is occasionally a missing current year’s reported revenue number due to the absence of a regulatory mandate for private companies to publicly disclose financial data. VentureSource relies in part on voluntary disclosure by private companies in their financial databases and also on regulatory disclosures. (S-1 filings associated with an IPO are a key data source. These S-1 filings can be used to ‘back-fill’ numbers not previously publicly disclosed.)
13 Beaver (1989: 91–101), Beaver and Morse (1978), and Foster (1986: 436–443) provide further discus-sion of the ‘transitory’ versus ‘permanent’ distinction with respect to reported earnings.
14 Specifically, for each sample of companies, we form 10 equally sized portfolios based on the observa-tion’s revenue growth from year t to year t + 1 (excluding the missing revenue growth observations). We then use the lowest and highest revenue growth for each portfolio to establish the revenue growth range of each of the 10 portfolios. The observations with missing revenue growth from year t – 1 to year t are then assigned to one of the 10 portfolios based on their revenue growth from year t to year t + 1. We report these assignments in Table 10.
15 Collins et al. (1999) report results showing an ‘anomalous significantly negative price-earnings relation using the simple earnings capitalization model for firms that report losses’ (p.29).
16 Due to possible collinearity among the company size, revenue growth, and profitability variables, the sum of these percentages may overstate the combined explanatory power of these variables in a multi-variate context.
17 The proposition that the percentage of losses has increased over time is supported by a univariate regres-sion of the percentage loss on a time trend. The trend term is positive and significant. Fama and French (2004) report evidence of an increase in loss rates for newly listed companies.
References
Allee KD and Yohn TL (2009) The demand for financial statements in an unregulated environment: an examination of the production and use of financial statements by privately held small businesses. The Accounting Review 84(1): 1-25.
Altman EI (2000) Predicting Financial Distress of Companies Revisiting the Z-Score and Zeta® Models, Working Paper, New York University.
Armstrong C, Davila A and Foster G (2006) Venture-backed private equity valuation and financial statement information. Review of Accounting Studies 17: 119-154.
Armstrong C, Davila A, Foster G and Hand JRM (2006) Biases in multi-year management financial forecasts: evidence from private venture-backed U.S. companies. Review of Accounting Studies 12(2–3): 183-215.
Ball R and Foster G (1982) Corporate financial reporting: a methodological review of empirical research. Journal of Accounting Research 20: 161-234.
Ball R and Watts R (1972) Some time series properties of accounting income. The Journal of Finance 27(3): 663-681.
Bartov E, Mohanram P and Seethamraju C (2002) Valuation of internet stocks—an IPO perspective. Journal of Accounting Research 40(2): 321-346.
Bauman CC, Bauman MP and Das S (2010) Valuation consequences of regulatory changes in revenue recog-nition: evidence from advertising barter sales. Advances in Accounting 26(2): 177-184.
Beaver WH (1989) Financial Reporting: An Accounting Revolution. 2nd ed. Englewood Cliffs, NJ: Prentice-Hall. Beaver WH and Morse D (1978) What determines price-earnings ratios? Financial Analysts Journal 34(4):
65-79.Berger PG (2003) Discussion of differential market reactions to revenue and expense surprises. Review of
Accounting Studies 8: 213-220.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
54 Australian Journal of Management 36(1)
Black F and Scholes M (1973) The pricing of options and corporate liabilities. Journal of Political Economy 81(3): 637-654.
Bowen RM, Davis AK and Rajgopal S (2002) Determinants of revenue reporting practices for internet firms. Contemporary Accounting Research 19: 523-562.
Callen JL, Robb SWG and Segal D (2004) Revenue Manipulation and Restatements by Loss Firms, Working Paper, University of Toronto.
Cassar G (2009) Financial statement and projection preparation in start-up ventures. The Accounting Review 84(1): 27-51.
Collins DM, Pincus M and Xie H (1999) Equity valuation and negative earnings: the role of book value of equity. The Accounting Review 74: 29-62.
Core JE, Guay WR and Van Buskirk A (2003) Market valuations in the new economy. Journal of Accounting & Economics 34: 43-67.
Davila A (2005) An exploratory study on the emergence of management control systems: formalizing human resources in small growing firms. Accounting, Organizations and Society 30(3): 223-248.
Davila A and Foster G (2005) Management accounting systems adoption decisions: evidence and perfor-mance implications from startup companies. The Accounting Review 80: 1039-1068.
Davila A and Foster G (2007) Management control systems in early-stage startup companies. The Accounting Review 82(4): 907-937.
Davila A, Foster G and Gupta M (2003) Venture capital financing and the growth of startup companies. Journal of Business Venturing 18: 689-708.
Davila A, Foster, G and Jia N (2010) Building sustainable high growth startup companies: management sys-tems as an accelerator. California Management Review 52(3): 79-105.
Davila A, Foster G and Li M (2009) Reasons for management control system adoption: insights from prod-uct development choice by early stage entrepreneurial companies. Accounting Organizations and Society 34(3–4): 322-247.
Davis A (2002) The value relevance of revenue for Internet firms: does reporting grossed-up or barter revenue make a difference. Journal of Accounting Research 40(2): 445-484.
Demers E and Lev B (2000) A rude awakening: internet shakeout in 2000. Review of Accounting Studies 6: 331-359.
Demers E and Lewellen K (2003) The marketing role of IPOs: evidence from Internet stocks. Journal of Financial Economics 68: 413-437.
Ertimur Y, Livnat J and Martikainen M (2003) Differential market responses to revenue and expense sur-prises. Review of Accounting Studies 8: 185-211.
Ertimur Y, Mayew WJ and Stubben SR (2011) Analyst reputation and the issuance of disaggregated earnings forecasts to I/B/E/S. Review of Accounting Studies forthcoming.
Fama EF and French KR (2004) New lists: fundamentals and survival rates. Journal of Financial Economics 73: 229-269.
Fama EF and Miller MH (1972) The Theory of Finance. New York: Holt, Rinehart and Winston, Inc.Foster G (1986) Financial Statement Analysis. 2nd ed. Englewood Cliffs, NJ: Prentice-Hall. Foster G, Davila A, Haemmig M, He X, Jia N, von Bismarck M, et al. (2011) Global Entrepreneurship and the
Successful Growth Strategies of Early-Stage Companies. World Economic Forum USA Inc.Ghosh A, Gu Z and Jain P (2005) Sustained revenue and earnings growth, earning quality, and earnings
response coefficients. Review of Accounting Studies 10: 33-57.Hand JRM (2001) The role of book income, web traffic, and supply and demand in the pricing of U.S. internet
stocks. European Finance Review 5: 295-317.Hand JRM (2003) Profits, losses and the nonlinear pricing of internet stocks. In: Hand JRM and Lev B (eds)
Intangible Assets: Values, Measures and Risks. New York: Oxford University Press.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 55
Hand JRM (2005) The value relevance of financial statements in the venture capital market. The Accounting Review 80: 613-648.
Hand JRM (2011) What drives the top line? Determinants of sales revenue in private venture-backed firms. In: Cumming D (ed.) Entrepreneurial Finance Handbook. New York: Oxford University Press.
Hayn C (1995) The information content of losses. Journal of Accounting and Economics 20: 125-153.Joos P and Plesko GA (2005) Valuing loss firms. The Accounting Review 80: 847-870.Keating EK, Lys TZ and Magee RP (2003) Internet downturn: finding valuation factors in spring 2000.
Journal of Accounting & Economics 34: 189-236.Keung EC (2010) Do supplementary sales forecasts increase the credibility of financial analysts’ earnings
forecasts. The Accounting Review 86(6): 2047-2074.Liu J, Nissim D and Thomas J (2002) Equity valuation using multiples. Journal of Accounting Research
40(1): 135-172.Ljungqvist A and Wilhelm WJ (2003) IPO pricing in the dot-com bubble. Journal of Finance 58(2): 723-752.Loughran T and Ritter J (2004) Why has IPO underpricing changed over time? Financial Management 33(3):
5-37.Miller MH and Modigliani F (1961) Dividend policy, growth and the valuation of shares. Journal of Business
34(4): 411-433.Ofek E and Richardson M (2002) The valuation and market rationality of internet stock prices. Oxford Review
of Economic Policy 18: 265-287.Ofek E and Richardson M (2003) DotCom mania: the rise and fall of internet stock prices. Journal of Finance
58(3): 1113-1138.O’Hara S and Lowry V (2009) Paying attention to sales: how market-cap-to-revenue makes a difference.
MarketWatch.Palepu KG and Healy PM (2007) Business Analysis and Valuation Using Financial Statements, Text and
Cases. Cincinnati, OH: South-Western.Penman S (2009) Financial Statement Analysis and Security Valuation. 4th ed. New York: McGraw-Hill/Irwin.Rajgopal S, Venkatachalam M and Kotha S (2002) Managerial actions, stock returns, and earnings: the case
of business-to-business Internet firms. Journal of Accounting Research 40(2): 529-556.Sandino T (2007) Introducing the first management control systems: evidence from the retail sector. The
Accounting Review 82(1): 265-293.Schultz P and Zaman M (2001) Do the individuals closest to internet firms believe they are overvalued?
Journal of Financial Economics 59: 347-381.Stubben SR (2010) Discretionary revenues as a measurement of earnings management. The Accounting
Review 85(2): 695-717.Trueman B, Wong MHF and Zhang XJ (2000) The eyeballs have it: searching for the value in internet stocks.
Journal of Accounting Research 38: 137-162.Trueman B, Wong MHF and Zhang XJ (2001) Back to basics: forecasting the revenues of internet firms.
Review of Accounting Studies 6: 305-329.Trueman B, Wong MHF and Zhang XJ (2003) Anomalous stock returns around internet firms’ earnings
announcements. Journal of Accounting & Economics 34: 249-271.
Appendix A
A.1 Risk differences across NYSE/AMEX/NASDAQ subsamples of listed companiesThe selected SIC industries and all other companies subsamples of sample one exhibit sizable dif-ferences in risk-related indicators prior to and after, as well as during, the 1998–2001 period.
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
56 Australian Journal of Management 36(1)
A.1.1 Capital market risk. Two frequently used measures of a company’s capital market risk are the Capital Asset Pricing Model (CAPM) beta and the standard deviation of security returns. The selected SIC industries group has had higher relative market risk and higher total market risk than the all other companies group for an extended time period since the 1980s through to 2004 (see Table A.1).
These data highlight that before, during, and after the 1998–2001 period, our selected SIC industry group had sizably higher market relative risk and higher market total risk than our all other companies group. With this higher risk that finance theory would predict came the potential for larger increases in market capitalization and larger decreases in market capitalization.
A.1.2 Negative net income. A fundamental indicator of company risk is the likelihood of reporting a loss. All else held equal, a firm with negative net income is less likely to generate funds for investing in new growth opportunities or to make distributions to its shareholders. The selected SIC industries group differs markedly from the all other companies group in terms of its propensity for losses. While there has been an increase since 1980 in the percentage of all companies reporting negative net income (with a decline in the 2002–2004 period), this increase has been concentrated in our selected SIC industries group (see Table A.2).
Table A.1. Capital market risk differences in selected years
Year Beta Standard deviation of returns
Selected SIC industries
All other companies
Selected SIC industries
All other companies
90th 50th 90th 50th 90th 50th 90th 50th
1980 1.87 1.08 1.54 0.70 0.044 0.029 0.042 0.0241985 2.15 0.87 1.42 0.55 0.041 0.022 0.036 0.0181990 1.77 0.68 1.36 0.42 0.072 0.029 0.055 0.0211995 2.09 0.75 1.33 0.35 0.053 0.024 0.037 0.0162000 1.71 0.86 1.08 0.24 0.086 0.040 0.054 0.0252004 2.14 1.17 1.63 0.71 0.049 0.023 0.031 0.014
SIC: standard industrial classification
Table A.2. Negative net income %'s for sample one in selected years
Year Selected SIC industries All other industries All companies pooled
1980 7.8% 23.5% 21.5%1985 30.3% 16.3% 18.5%1990 37.2% 23.1% 25.6%1995 39.1% 17.2% 21.9%2000 56.8% 22.3% 31.0%2001 68.0% 25.3% 36.0%2004 50.3% 22.7% 29.4%
SIC: standard industrial classification
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from
Armstrong et al. 57
The results for the all companies column have been noted before by Collins et al. (1999), Hayn (1995), and Joos and Plesko (2005). Over time, the likelihood that a publicly traded U.S. listed company will report a loss has increased. In 1980, there was roughly a 20% chance in any one year, while by 2000 there was more than a 30% chance.17
Companies on the NASDAQ have higher loss percentages than companies on the NYSE/AMEX. Selected percentages over the 1980–2000 period are given in Table A.3.
NASDAQ companies are, on average, smaller and younger than companies on the NYSE/AMEX. Size and age have been found to be useful predictors of financial distress (Altman, 2000).
Date of acceptance of final transcript: 3/2/2011.Accepted by Associate Editor, Peter Clarkson (Accounting).
Table A.3. Negative net income %'s for NASDAQ and NYSE/AMEX in selected years
Year NASDAQ companies NYSE/AMEX companies All companies
1980 23.6% 21.3% 21.5%1985 21.5% 16.7% 18.5%1990 31.5% 21.5% 25.6%1995 27.5% 15.7% 21.9%2000 40.5% 18.3% 31.0%2001 44.1% 25.5% 36.0%2004 36.4% 20.3% 29.4%
NSYE: New York Stock Exchange, AMEX: American Stock Exchange, NASDAQ: National Association of Securities Deal-ers Automated Quotation System
at UNIV OF COLORADO LIBRARIES on February 16, 2012aum.sagepub.comDownloaded from